-------------------------------------------------------------------------------
Matrox Imaging Library (MIL) X Service Pack 4
Release Notes (whatsnew_processing)
March 2020
(c) Copyright Matrox Electronic Systems Ltd., 1992-2020.
-------------------------------------------------------------------------------
Main table of contents
Section 1 : Differences between MIL X Service Pack 4 and MIL 10
Processing Pack 3
Section 2 : Differences between MIL 10 Processing Pack 3 and MIL 10
Processing Pack 2 with Update 63
Section 3 : Differences between MIL 10 Processing Pack 2 with Update 63 and
MIL 10 Processing Pack 2
Section 4 : Differences between MIL 10 Processing Pack 2 and MIL 10
Processing Pack 1
Section 5 : Differences between MIL 10 Processing Pack 1 and MIL 10
Section 6 : Differences between MIL 10 and MIL 9 Processing Pack 2 with
Update 56
Section 7 : Differences between MIL 9 Processing Pack 2 with Update 56 and
MIL 9 Processing Pack 2
Section 8 : Differences between MIL 9 Processing Pack 2 with Update 45 and
MIL 9 Processing Pack 2
Section 9 : Differences between MIL 9 Processing Pack 2
and MIL 9 Processing Pack 1
Section 10 : Differences between MIL 9 Processing Pack 1 and MIL 9
Section 11 : Differences between MIL 9 and MIL 8 Processing Pack 4
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 1: Differences between MIL X Service Pack 4 and MIL 10
Processing Pack 3
Table of Contents for Section 1
1. Overview
2. New functionalities and improvements
2.01 MIL processing specific examples
2.02 Classification module
2.03 Blob module
2.04 Calibration module
2.05 SureDotOCR® module
2.06 Registration module
2.07 Pattern Matching module
2.08 Model Finder module
2.09 Metrology module
2.10 3dMap module
2.11 Code Reader module
2.12 Measurement module
2.13 String Reader module
2.14 Primitives
2.15 Graphics
2.16 3D metrology module
2.17 3D registration module
2.18 3D image processing
2.19 3D geometry module
2.20 3D graphic list
2.21 General
3. Deprecated functionalities
3.01 Classification module
3.02 Calibration module
3.03 3dMap module
3.04 Code Reader module
3.05 Primitives
3.06 General
4. Fixed bugs
4.01 Classification module
4.02 Blob module
4.03 Calibration module
4.04 SureDotOCR® module
4.05 Edge Finder module
4.06 Registration module
4.07 Pattern Matching module
4.08 Model Finder module
4.09 Metrology module
4.10 Code Reader module
4.11 String Reader module
4.12 Primitives
4.13 Graphics
5. Known limitations
5.01 Classification module
5.02 String Reader module
5.03 3D display
-------------------------------------------------------------------------------
1. Overview
- MIL X Service Pack 4 includes all the features of MIL 10
Processing Pack 3.
In addition, MIL X Service Pack 4 includes new processing
functionalities, performance optimizations, and general improvements,
such as:
. New: Training of image-oriented CNN classifier using predefined
architectures.
. New: Detection using image-oriented CNN classifier.
. New: Feature-oriented Tree Ensemble classifier (training and
predicting).
. New: MclassPredict speed optimization for classification and
detection.
. New: MimAugment to perform data augmentation for classification among
other uses.
. New: 3D API overhaul; some M3dmap functionality has moved to new
3D modules (M3dim, M3dgeo, M3dmet and M3dreg).
. New: Additional 3D functionalities, including point cloud cropping,
merging, sampling and meshing, cylinder and 3D line fitting,
registration of multiple point clouds, distance computation
between point clouds, depth maps and fitted surfaces.
. New: 3D transformation matrix manipulation.
. New: 3D display of depth maps, point clouds, and analytic or fitted
3D shapes supporting LUT coloring, opacity and line thickness.
. New: Container data object useful for, among others, 3D acquisition
and processing.
. New: MbufConvert3d to convert most GenICam PFNC/SFNC 3D pixel
formats/coordinate system output modes to processable point
clouds.
. New: Code Reader support for Extended Rectangular Data Matrix (DMRE).
. New: Code Reader support for grading according to ISO/IEC 15416:2016.
. New: Code Reader support for grading according to Semi T10-0701.
. New: Additional processing primitives such as function to perform
differential operator, to compute bounding box, etc.
. New: Several primitives accelerated using Intel AVX-512 instruction
set.
. New: Numerous new and updated examples.
. Many more additions and improvements to discover!
- MIL 10 is the minimum requirement for all upcoming Service Packs
until the next major release.
- MIL X Service Pack 4 does not support Windows 8.1. It requires
Windows 7 or Windows 10 1607 (or a more recent Windows 10 build).
- The training of a CNN classifier requires Windows 10 1607 64-bit
(or a more recent Windows 10 64-bit build).
- MIL X Service Pack 4 does not support the Matrox CronosPlus,
Matrox Iris GT, Matrox Radient eV-CLHS, Matrox Vio, GPU and
IEEE 1394 IIDC systems.
- Windows’ automatic 8.3 file name creation needs to be enabled in order
for the MIL installer to access the temp folder when the user name
contains a space. This option allows Windows to create short file/folder
name aliases for ones with long names for programs, such as the MIL
installer, that don't support spaces in the file/folder names.
Alternatively, the MIL installer needs to run from a user account that
belongs to the administrators group and has no spaces in it. Note that the
same applies for uninstalling MIL.
- Some anti-virus or security applications may interfere with your MIL
installation. In order to perform the installation without any
interruption, it is recommended that you temporarily disable your
anti-virus or security application.
- The required Visual C++ 2017 Redistributable also needs the presence of
KB2919442 and KB2919355. These will need to be obtained and applied to
install this Service Pack.
- Building the MIL examples using Visual Studio 2015 or 2017 requires
the presence of Windows SDK version 8.1, which is installed from the
Visual Studio setup.
- Shortcuts to the ModelFinder and EdgeFinder interactive utilities have been
removed from MIL because the functionality they provide has been
incorporated in the MIL CoPilot interactive environment included in the
latest MIL release or available as an update to MIL.
- Ply files are associated to Microsoft’s 3D Viewer by default on Windows 10.
To open a ply file with 3D Viewer, the point cloud must contain a mesh;
otherwise, you will get the error: “Couldn’t load 3D model”.
- If you have installed and prototyped using MIL X Service Pack 4 Early
Access, you may need to recompile your application using
MIL X Service Pack 4.
- MIL CoPilot from Update 94 Early Access may not be fully compatible with
MIL X Service Pack 4 Official Release. It is strongly recommended
to update MIL CoPilot with Update 94 Official Release as soon as it
becomes available.
- MIL X Service Pack 4 will be the last version with Python 2.7 support.
- The container buffer support for the GenTL Consumer provided in MIL 10
Update 34 and MIL X Version 1911 is incompatible with the container buffer
support in MIL X Service Pack 4. Calling MbufAllocDefault() for a
M_SYSTEM_GENTL system with the M_CONTAINER attribute will generate an
error. Moreover, the MultiComponentGrab example, which uses containers,
will also generate an error. The issue will be resolved in a future update
to the GenTL Consumer.
2. New functionalities and improvements
2.01 MIL processing specific examples
- New MIL X Service Pack 4 examples!
- Note that MIL X Update 81 is needed for the images used by several
examples, in particular for the 3D processing ones.
- Online repository for third-party 3D camera interface examples (to come).
- More examples to come in the official release of MIL X Service Pack 4.
2.02 Classification module
CNN CLASSIFIER
- New: MclassTrain() allows you to perform CNN training from the ground up
using the CPU or supported GPU. Specifically, you can perform a complete
training, as well as transfer learning and fine tuning.
- New: The training can be parametrized using MclassControl on an
M_TRAIN_CNN context.
- New: Five predefined CNN MIL classifier contexts are provided to be
trained with MclassTrain(). They are located in the MIL installation path
in the "Contexts" folder. They address different situations. These
contexts can be allocated and loaded in one call, using MclassAlloc with
M_FCNET_S, M_FCNET_M, M_FCNET_XL, M_FCNET_MONO_XL or M_FCNET_COLOR_XL.
- New: MclassAlloc(..., M_DATASET_IMAGES, ...) allows you to allocate MIL
dataset objects that are mandatory to train a CNN classifier using
MclassTrain(). These MIL dataset objects are built to handle the pairings
between the images and associated classes.
- New: MclassControl(..., M_ENTRY_ADD, ...) allows you to add entries to a
dataset.
- New: MclassControlEntry() allows you to set information to an entry of the
dataset.
- New: MclassSplitDataset() allows you to split a source dataset context into
two destination dataset contexts. This is useful to split your data into a
train and development dataset, both of which are required by MclassTrain().
- New: MclassHookFunction() allows you to attach or detach a user-defined
function to a CNN training event (M_MINI_BATCH_TRAINED and/or
M_EPOCH_TRAINED).
- New: MclassGetHookInfo() allows you to retrieve useful information inside a
user-defined hook function.
- New: MclassCopyResult(..., M_TRAINED_CLASSIFIER_CNN, ...) allows you to copy
a trained context from a train result to a M_CLASSIFIER_CNN_PREDEFINED
context.
- New: MclassImport() allows you to import information from a csv file to a MIL
dataset. Information can be entries, class definitions, etc.
- New: MclassExport() allows you to export information from a MIL dataset to a
csv file. Information can be entries, class definitions, etc.
It also allows you to export training results to a csv file.
- Improvement: MclassPredict() has been further optimized. Speed
optimizations range from 2x faster to almost 10x faster.
- New: MclassPredict() now natively supports Coarse Segmentation (i.e.
classification using a target image that is larger than the CNN's source
layer and producing a classification map).
- New: MclassDraw() allows you to draw specific features of prediction
results. It also allows you to draw class definition sample images.
- New: M_CLASS_DRAW_COLOR and M_CLASS_NAME controls/inquires are now
available for the class definition sub-objects in a classification context.
- New: To validate that the current system meets the requirements to perform
optimal CNN training, a CNN Train Engine Test is available in the
MILConfig tool under the Classification folder.
TREE ENSEMBLE CLASSIFIER
New classifier.
- New: MclassTrain() allows you to perform tree ensemble training from the
ground up or from an already trained classifier (warm start).
- New: The training can be parametrized using MclassControl on a
M_TRAIN_TREE_ENSEMBLE context.
- New: MclassPredict() allows you to predict the category of an input feature
vector (array) using a trained tree ensemble classifier.
- New: MclassAlloc(..., M_DATASET_FEATURES, ...) allows you to allocate MIL
dataset objects that are mandatory to train a tree ensemble classifier
using MclassTrain(). The MIL dataset object is built to handle the pairings
between the features and associated classes.
- New: MclassControl(..., M_ENTRY_ADD, ...) allows you to add entries to a
dataset.
- New: MclassControlEntry() allows you to set information to an entry of the
dataset.
- New: MclassSplitDataset() allows you to split a source dataset context into
two destination dataset contexts. This is useful to split your data into a
train dataset and development dataset.
- New: MclassHookFunction() allows you to attach or detach a user-defined
function to a tree ensemble training event (M_TREE_TRAINED).
- New: MclassGetHookInfo() allows you to retrieve useful information inside a
user-defined hook function.
- New: MclassCopyResult(..., M_TRAINED_TREE_ENSEMBLE, ...) allows you to copy
a trained context from a train result to a M_CLASSIFIER_TREE_ENSEMBLE
context.
- New: MclassImport() allows you to import information from a csv file to a MIL
dataset. Information can be entries, class definitions, etc.
- New: MclassExport() allows you to export information from a MIL dataset to a
csv file. Information can be entries, class definitions, etc.
It also allows you to export training results to a csv file.
- New: M_CLASS_DRAW_COLOR and M_CLASS_NAME controls/inquires are now
available for the class definition sub-objects in a classification context.
2.03 Blob module
- Improvement: MblobDraw drawing of M_FERET_MIN_DIAMETER and
M_FERET_MAX_DIAMETER using contact points.
2.04 Calibration module
- New: McalFixture now accepts M3dgeo transformation matrix as a destination
object to which the calibration fixture matrix is written.
- New: McalGetCoordinateSystem and McalSetCoordinateSystem now supports a
M3dgeo matrix.
- New: McalTransformCoordinate3dList() now supports 32-bit unsigned buffers
as depth maps. This is also true for 3dim and 3dmet functions which take
depth maps as input.
2.05 SureDotOCR® module
- New: M_STRING_PITCH, M_STRING_PITCH_MODE, M_ITALIC_PITCH, M_ITALIC_PITCH_MODE
can now be used to control the distance between dot centers in the string or
italic angle direction using MdmrControl().
- New: MdmrGetResult() has new results M_STRING_PITCH, M_STRING_CHAR_ANGLE,
M_ITALIC_PITCH and M_ITALIC_CHAR_ANGLE to retrieve the string and italic
pitches and angles used to read the characters.
- Improvement: MdmrRead() has been improved when M_STRING_ANGLE_MODE is set
to M_ANGLE or M_ORIENTATION. It will now less likely mismatch characters,
which occurred in rare cases.
- Improvement: MdmrRead() has been improved and will now less likely fail
because of incorrect orientation or pitch estimation, which occurred in
rare cases.
- Improvement: MdmrRead() now makes better use of multiprocessing
capabilities.
2.06 Registration module
- New: MregCalculate() now supports HDR (High Dynamic Range) functionality.
- New: MregControl() now supports M_DRAW_REMAP_FACTOR_MODE and
M_DRAW_REMAP_FACTOR_VALUE to control the drawing remap of
M_GAUSSIAN_CURVATURE_IMAGE, M_MEAN_CURVATURE and M_LOCAL_SHAPE.
- New: MregGetResult() now supports M_RANGE_FACTOR_GAUSSIAN_CURVATURE,
M_RANGE_FACTOR_MEAN_CURVATURE and M_RANGE_FACTOR_LOCAL_SHAPE values.
- Improvement: MregCalculate() with M_PHOTOMETRIC_STEREO now has a better
responsive M_TIMEOUT.
- Improvement : Improved numerical results for Local Shape image result
(M_DRAW_LOCAL_SHAPE_IMAGE) when light vectors' azimuth were not aligned
with the image axis.
2.07 Pattern Matching module
- New: MpatFind() now supports models of different size when M_SEARCH_MODE
is set to M_FIND_ALL_MODELS (or M_DEFAULT).
- New: MpatDefine() can now use a source image associated with a graphic
list to define a model from a rotated bounding box.
2.08 Model Finder module
- New: MmodDefine() can now use a source image associated with a graphic
list to define a model from a rotated bounding box.
- New: MmodGetResult() can now return corner positions of a rectangle result
occurrence with M_SHAPE_RECTANGLE result, using the result types:
M_TOP_LEFT_X, M_TOP_LEFT_Y, M_TOP_RIGHT_X, M_TOP_RIGHT_Y, M_BOTTOM_LEFT_X,
M_BOTTOM_LEFT_Y, M_BOTTOM_RIGHT_X, and M_BOTTOM_RIGHT_Y.
- New: MmodFind() now supports search regions at an angle for M_SHAPE_CIRCLE,
M_SHAPE_ELLIPSE, M_SHAPE_RECTANGLE and M_SHAPE_SEGMENT. Regions at an angle
can be set using MmodControl(..M_ANGLE_REGION..), using
MmodControl(..M_SEARCH_POSITION_FROM_GRAPHIC_LIST..), or using a target
image containing a rectangular region of interest set with MbufSetRegion().
2.09 Metrology module
- New: MmetPut() now supports MIL_FLOAT arrays for positions and angles.
- Improvement: MmetSetRegion(), MmetAddTolerance() and MmetControl() have
been accelerated when numerous features are present in the metrology
context.
- New: MmetDraw() now supports M_DRAW_TOLERANCE_AREA as an operation to draw
the area used to compute M_AREA_BETWEEN_CURVES, M_AREA_CONVEX_HULL,
M_AREA_SIMPLE, M_AREA_UNDER_CURVE_MAX and M_AREA_UNDER_CURVE_MIN.
- Improvement: MmetDraw() with Operation set to M_DRAW_TOLERANCE has been
improved for area and perimeter tolerances.
2.10 3dMap module
- New: M3dmapCopyResult() to copy a group of results from a 3D
reconstruction result buffer into a container or an image buffer.
- New: M3dmapDraw3d() to draw 3D annotations based on a laser line profiling
3D reconstruction context.
2.11 Code Reader module
- Improvement: McodeRead(), McodeGrade() and McodeTrain() have been
accelerated for Qr code type.
- Improvement: Robustness of M_IMPROVED_RECOGNITION initialization mode has
been improved. The M_ENCODING is now set to M_ANY for all code types
that support M_ANY as a valid value. M_POSITION_ACCURACY is set to M_HIGH.
- New: M_DOT_SPACING_INPUT_UNITS, M_DOT_SPACING_MIN and M_DOT_SPACING_MAX
are now available for M_MICROQRCODE and M_MAXICODE code types.
- Improvement: McodeTrain() has been improved to train 2D code types with
small cell size.
- Improvement: McodeRead() now supports erasures when using M_MICROPDF417
code type. This results in a more robust read operation in some situations.
- Improvement: The localization used for linear bar codes in M_VERY_LOW speed
and in multi occurrence use cases has been improved. It is more robust
(reads more codes) and faster in most images.
- New: Extended Rectangular Data Matrix (DMRE) is now supported by
McodeRead(), McodeWrite(), McodeTrain() and McodeGrade(). It has been
integrated to the M_DATAMATRIX code type.
- Improvement: McodeRead()'s ability to decode 1D bar codes in blurry and
noisy images has been improved.
- Improvement: McodeRead()’s ability to decode EAN/UPC bar codes in low
quality images (noise, bad modulation, eroded/blurred bars) has been
improved.
- Improvement: McodeRead()’s ability to decode 1D bar codes with small cell
size (<= 2 pixels) and at angles that are multiple of 90 degrees has been
improved. The grading results are also now more accurate in this
situation.
- Improvement: McodeRead()’s speed has been improved when reading a linear
code with M_SPEED set to M_VERY_LOW and when another linear code is
horizontally aligned with the one to read.
- New: It is now possible to retrieve the score of each decoded scan for the
linear bar code and composite code after an McodeRead() or McodeGrade()
operation. It is also possible to draw those decoded scans using
McodeDraw().
- New: M_ELEMENT_NUMBER_X and M_ELEMENT_NUMBER_Y results are now available
after an McodeWrite(), McodeRead() and McodeGrade() operation.
M_ELEMENT_NUMBER_X is useful to retrieve the number of elements
(bars + spaces) for linear bar codes.
- Improvement: McodeRead() robustness has been improved for EAN/UPC codes.
Avoidable fail or wrongly decoded string due to mixing the pairs 1,7 or
2,8 are now very rarely.
- Improvement: McodeTrain() will now return more accurate recommended
M_DOT_SPACING_MIN and M_DOT_SPACING_MAX values.
- New: McodeStream() with operation set to M_SAVE_REPORT can now be used to
save a report containing most results from a McodeTrain() operation.
- Improvement: Robustness of M_IMPROVED_RECOGNITION initialization mode has
been improved for all linear bar codes except M_PHARMACODE. By default,
M_THRESHOLD_MODE is now set to M_GLOBAL_SEGMENTATION.
To use the old behavior, the user must call
McodeControl(ContextCodeId, THRESHOLD_MODE, M_ADAPTIVE).
- New: It is now possible to control the cell size of a code for an
McodeWrite() operation using M_CELL_SIZE_MODE (M_AUTO or M_USER_DEFINED).
When the mode is set to M_USER_DEFINED, the cell size can be controlled
using M_CELL_SIZE_VALUE. Note that the legacy way to specify the cell size
for an McodeWrite using M_CELL_SIZE_MIN is not functional anymore.
- New: More results are now available after an McodeTrain() operation
(M_TRAIN_ENABLED_CONTROL_TYPES, M_TRAIN_ENABLED_CONTROL_TYPES_STATE,
M_TRAIN_ENABLED_CONTROL_TYPES_ORIGINAL_VALUE,
M_TRAIN_ENABLED_CONTROL_TYPES_TRAINED_VALUE, and
M_TRAIN_ENABLED_NUMBER_OF_MODEL_CONTROL_TYPES). These are useful to
retrieve information about the controls that were enabled to be trained
(control type, training state, original value and trained value).
- New: McodeTrain() now permits to retrieve the list of controls that will
be changed by a call to McodeControl(..., M_RESET_FROM_TRAINED_RESULTS,...).
The list can be retrieved using McodeGetResult(M_TRAINED_CONTROL_TYPES).
It is possible to retrieve the list for the context or a code model
depending on the ResultIndex used.
- Improvement: In a DMIL configuration with a code context Id and grade result
Id allocated on two different systems, the error message sent to the user
will now explicitly mention that this type of configuration is not allowed.
- New: McodeModel() with operation value M_ADD + M_SUPPORTED can now be used
to know if a code type could be added to an existing context.
- New: McodeDraw() now supports operations M_DRAW_BOX, M_DRAW_POSITION and
M_DRAW_QUIET_ZONE after an McodeWrite() operation.
- New: McodeWrite() can now encode a sample string when the parameter
StringPtr is set to M_NULL.
- New: McodeWrite() can now generate dot printed codes for Datamatrix,
QR Code, Micro QR Code and Aztec code types using the new control parameter
M_DOT_SHAPE and setting it to M_CIRCLE. It is also now possible to control
the size of the dot itself independently of the cell size (the distance
between the center of two cells) using the new control parameter M_DOT_SIZE.
- New: McodeGrade() can grade according to ISO/IEC 15416:2016. This edition
is used by default. To grade according to a specific standard or edition,
use the new model control type M_GRADING_STANDARD_EDITION.
Possible values of M_GRADING_STANDARD_EDITION are:
.M_DEFAULT (same as most recent supported grade edition).
.M_ISO_15416_2000 (for 1D model type).
.M_ISO_15416_2016 (for 1D model type).
.M_ISO_15415_2011_15416_2000 (for 2D matrix, cross row and composite code
model type).
.M_ISO_15415_2011_15416_2016 (for 2D matrix, cross row and composite code
model type).
.M_ISO_29158_2011 (for 2D matrix).
- New: McodeGrade() can grade Datamatrix according to Semi T10-0701
standards.
- Improvement: McodeRead() will now be faster when reading a GS1 Databar or
composite code of sub-type M_GS1_DATABAR_STACKED or
M_GS1_DATABAR_STACKED_OMNI and when speed is set to M_HIGH.
- Improvement: McodeDraw(M_DRAW_BOX) angle precision has been improved for
the 2D part of a composite code.
- Improvement: McodeRead() is now less likely to miss some occurrences when
reading in images containing multiple 1D codes (was happening in rare cases).
- Improvement: McodeRead() is now less likely to perform a short read on 1D
bar codes when speed is set to very low (was happening in rare cases).
- New: McodeDraw() now supports the operations M_DRAW_CODE and M_DRAW_POSITION
after an McodeGrade().
- Improvement: The trained control type settings produced by McodeTrain()
will result in a better optimized context in some situations.
- New: McodeModel(...M_RESET_FROM_DETECTED_RESULTS...) can now be used to
reset models in a context using results from McodeDetect().
- Improvement: The accuracy of the M_ANGLE result has been improved when
reading 1D bar codes in some situations.
- New: The M_CODE_TYPE result can now be retrieved for each code model
trained after a call to McodeTrain().
- Improvement: Result types M_R_MIN and M_R_MAX are now available for all
code types that support an McodeGrade() operation. Before they were only
available for 2D matrix code types.
- Improvement: McodeGrade() has been accelerated for 2D Cross-row code types
and composite codes.
- Improvement: McodeRead() will no longer be slower when decoding a scanline
on the edge of a 1D code (was happening in rare cases).
2.12 Measurement module
- Improvement: MmeasFindMarker() numerical stability has improved when
setting M_SEARCH_REGION_CLIPPING to M_ENABLE. The results of an
MmeasFindMarker() operation will be less subject to change when updating
the MIL version or running under different OS (as was happening in rare
cases).
2.13 String Reader module
- Improvement: A MstrPreprocess() preprocessed context will now use up to
30% less MIL Ids and will require less memory.
2.14 Primitives
- New: MimAugment() performs an augmentation operation which allows you to
create a plausible variation of an image. This can be used for data
augmentation when training a classifier context (Mclass).
- New: MgenLutFunction() now supports M_COLORMAP_TURBO, which is an improved
version of M_COLORMAP_JET, with smoother color transitions.
- New: MgenLutFunction() now supports M_COLORMAP_DISTINCT_256, which is
designed to maximize the visual difference between consecutive colors.
- New: MgenLutFunction() now supports the new mathematical function
M_SIGMOID to fill the LUT.
- Improvement: MimDistance(), MimMatch(), MimEdgeDetect(),
MimWarp(M_WARP_LUT), MimLutMap(), MimConvolve() and MimBinarize() (integer
to binary) are speed-optimized using Intel AVX-512 instruction set.
- New: MimBoundingBox(), function to compute a bounding box for pixel values
meeting a specified condition in an image.
- New: MgenLutFunction() now supports M_TONE_MAPPING which allows you to
create a pseudo-gamma tone mapping function using an M_POWER function.
- New: MgenLutFonction() now supports M_FIRST_GRAY and M_LAST_GRAY options for
colormap.
- New: MimFilterAdaptive() with M_NOISE_PEAK_REMOVAL filter mode. This is a
simple and fast filter that removes salt and pepper noise.
- New: MbufChildColor(), MbufCopyColor(), MbufGetColor() and MbufPutColor()
functions support the bands M_HUE, M_SATURATION and M_V for HSV image.
- New: MimDifferential(), function to perform differential operator. The
available operators are: M_GRADIENT, M_GRADIENT_SQR, M_LAPLACIAN and
M_SHARPEN.
- New: MimConvolve(), with linear filter context, now supports a Vliet IIR
filter.
- New: MbufClone() now supports any valid MIL system identifier for the SysId
parameter. This will allocate the clone on that system.
- New: MbufClone() now supports any valid attribute for the Attribute
parameter. This will replace the attributes from the source buffer.
- New: MgenRamp(), function to generate a ramp image using scales in x and y
and an offset.
- Improvement: MbufClear() speed has been improved when a region of interest
is associated to the buffer to be cleared.
- New: MbufSetRegion() supports new operations: M_LINK_TO_PARENT, M_COPY and
M_EXTRACT.
- New: MbufControl() supports M_REGION_USE that can be set to M_USE or
M_IGNORE.
- New: M_REGION_ACTIVE can now be inquired using MbufInquire() to know if a
region is considered active by the MIL functions (active if M_REGION_TYPE
is different than M_NONE and M_REGION_USE is M_USE).
- Improvement: MbufAlloc1d() will no longer allocate padding pixels because
this was resulting in unnecessary memory usage.
- New: MbufInquire() now supports M_HOST_ADDRESS_BAND+n as an InquireType.
This will avoid having to create band child buffers just to inquire the
host addresses.
- New: MbufInquire() now supports the InquireTypes
M_EXTENDED_ATTRIBUTE_NAME and M_EXTENDED_FORMAT_NAME.
2.15 Graphics
- Improvement: Dashed lines in the selection box for interactive graphics
are now more responsive.
- Improvement: Flickering when hovering over editable graphics has been
reduced by eliminating some unnecessary display updates.
2.16 3D metrology module
- New: M3dmetFit() fits a plane or sphere geometry on a point cloud, and
replaces the deprecated M3dmapSetGeometry. It is now possible to use a
robust fit mode, to fit cylinder and line geometry objects. The fit
can be parametrized using a M_3DMET_FIT_CONTEXT.
- New: M3dmetDistance() calculates the distance between a point cloud and a
reference object, and replaces M3dmapDistance(). In addition, it calculates
the distance to sphere surface, cylinder surface or axis, line, box,
Z distance to a depth map, distance to mesh or points of another point
cloud.
- New: M3dmetStat() computes statistics on depth maps, and replaces the
deprecated M3dmapStat(). In addition to depth maps, it now accepts point
clouds as a source. The reference object can also be a box, a cylinder, or
a point cloud.
- New: M3dmetVolume() computes the volume of a depth map, with respect
to an optional reference plane, and replaces the deprecated M3dmapStat()
with M_VOLUME. In addition, computing the volume of a mesh and using
depth maps as reference are now supported.
2.17 3D registration module
- New: M3dreg module replaces M3dmapAlign(), with registration elements
support similar to 2D registration module.
- New: M3dregCalculate() performs the registration operation on the given
point clouds and writes the resulting transformation matrices into the
pairwise 3D registration result buffer. Previously, the deprecated
M3dmapAlign() would align two point clouds only. It also offers more
subsampling options.
- New: M3dregMerge() merges point clouds using the relations found in a
previous call to M3dregCalculate().
2.18 3D image processing
- New: M3dimTranslate() translates the 3D points in a container or 3D
geometry object.
- New: M3dimRotate() applies a rotation to the 3D points in a container or
3D geometry object.
- New: M3dimScale() scales the 3D points in a container or 3D geometry
object.
- New: M3dimMatrixTransform() applies the given transformation matrix to the
3D points in a container or 3D geometry object.
- New: M3dimMatrixTransformList() transforms a list of coordinates (3D
points) using the specified transformation matrix.
- New: M3dimCrop() crops points in a point cloud based on a given 3D
geometry or calibrated mask image buffer.
- New: M3dimSample() subsamples a point cloud to generate a less dense point
cloud, or samples a meshed point cloud to generate a denser point cloud.
- New: M3dimRemovePoints() removes points from a container based on a
given condition.
- New: M3dimProfile() creates a 3D profile from a depth map or point cloud.
- New: M3dimMerge() merges multiple point clouds into a single point cloud.
- New: M3dimNormals() computes a point cloud's unit normal vectors.
- New: M3dimMesh() creates a 3D mesh structure for a point cloud.
- New: M3dimCalculateMapSize() estimates a recommended depth map image
buffer size from a source container's point cloud.
- New: M3dimCalibrateDepthMap() calibrates a depth map image buffer based on
the bounding box of a container's point cloud or an axis-aligned 3D box
geometry object.
- New: M3dimProject() projects the 3D points in a container or a 3D geometry
object into a fully corrected depth map, and replaces the deprecated
M3dmapExtract(). In addition, an M_MESH_BASED projection mode is supported.
It can also project 3D geometry objects. A new M_AVERAGE_MODE overlap mode
is supported.
- New: M3dimFillGaps() fills gaps in a depth map and, optionally, an
intensity map, according to gap filling context controls, and replaces the
deprecated M3dmapExtract. In addition, a new fill border replicate mode is
supported.
- New: M3dimArith() performs point-to-point arithmetic operations on depth
maps and 3D geometries, and replaces M3dmapArith(). In addition, there is
a new M_REPLACE_MISSING_DATA operation, and support for line and cylinder
geometries.
- New: M3dimStat() computes a variety of statistics on depth maps, and
replaces the deprecated M3dmapStat(). In addition, point clouds are
supported and new statistics like centroid and distance to nearest
neighbor are now supported.
- New: M3dimCopy() copies a depth map calibration into a 3D geometry of type
M_BOX.
2.19 3D geometry module
To allocate, define and display 3D geometry objects or to allocate and
define transformation matrix objects.
- New: Line and cylinder geometry support.
- New: M3dgeoMatrix functions to create and compose 4x4 transformation
matrices.
2.20 3D graphics list
To allocate and populate a 3D graphics list with graphics and point clouds.
Useful to display 3D objects and point cloud or depth map on a 3D display.
2.21 General
- Improvement: MdigAlloc() is now faster for large sets of images when
InitFlag is set to M_EMULATED.
- New: MdigAlloc() now supports "M_2D_SIMULATOR" and "M_3D_SIMULATOR" with
a simulated digitizer.
Note: You must use the appropriate default digitizer for your application.
For example, if you specify M_3D_SIMULATOR, and you run an example that
uses MdigGrab() to acquire 2D images, you might obtain 1-pixel high images
with reinterpreted 3D data content. The default digitizer is set during
MIL installation and can be changed and using the MILConfig utility.
- New: MappFileOperation() now supports M_FILE_MOVE, M_FILE_NAME_FIND_COUNT,
and M_FILE_NAME_FIND operations.
- New: MappFileOperation() now supports the OperationFlag M_RECURSIVE for
M_FILE_DELETE_DIR to delete the requested directory and all its
subdirectories.
- New: MbufExport() now supports M_USE_EXTENSION as a file format.
- New: MobjInquire() now supports M_OBJECT_FILE_PATH, M_OBJECT_FILE_NAME,
M_OBJECT_FILE_NAME_NO_EXTENSION, M_OBJECT_FILE_EXTENSION and
M_OBJECT_FILE_FOLDER as InquireTypes.
- Improvement: MIL no longer reuses freed MIL_ID values before all possible
values have been used. This will help debugging some types of problems.
- New: MappInquire() can now be used to retrieve common MIL paths using the
InquireType M_MIL_DIRECTORY_INSTALL, M_MIL_DIRECTORY_EXAMPLES,
M_MIL_DIRECTORY_IMAGES and M_MIL_DIRECTORY_CONTEXTS.
- New: MIL_UNIQUE_..._ID can be used in C++ to manage the lifetime of a
MIL_ID. Use M_UNIQUE_ID in Alloc functions to get a MIL_UNIQUE_..._ID,
which will call the appropriate Free function automatically.
- New: M_TO_STRING() converts its argument to a MIL_STRING.
- New: MfuncAllocScript(M_INTERPRETER_C_PYTHON37) allocates a script-based
user-defined function for Python 3.7.
3. Deprecated functionalities
3.01 Classification module
- MclassControl() with M_PREDICT_MODE is deprecated, as are its two values
M_SINGLE_CLASSIFICATION and M_CLASSIFICATION_MAP.
The predict mode M_SINGLE_CLASSIFICATION is still functional and will be the
only available mode.
The predict mode M_CLASSIFICATION_MAP is replaced by the natively supported
coarse segmentation
- MclassControl() with M_CLASSIFICATION_MAP_INPUT_SIZE_X,
M_CLASSIFICATION_MAP_INPUT_SIZE_Y, M_STRIDE_X and M_STRIDE_Y are deprecated.
- M_PRETRAINED_CONVOLUTIONAL_NETWORK has been renamed to
M_CLASSIFIER_CNN_PREDEFINED.
- M_SAMPLE_IMAGE_ID has been renamed to M_CLASS_ICON_ID.
3.02 Calibration module
- Deprecation: M_POINT_CLOUD_CONTAINER_3DMAP is deprecated and replaced by
M_RESULT_POINT_CLOUD_3DMAP.
3.03 3dMap module
- Deprecation: M_POINT_CLOUD_CONTAINER is deprecated and replaced by
M_POINT_CLOUD_RESULT.
- Deprecation: M_3DMAP_POINT_CLOUD_CONTAINER is deprecated and replaced by
M_3DMAP_POINT_CLOUD_RESULT.
- Deprecation: M_POINT_CLOUD_CONTAINER_3DMAP is deprecated and replaced by
M_RESULT_POINT_CLOUD_3DMAP.
- Deprecation: M3dmapAlign() is deprecated and alignment can now be done
using the M3dregCalculate() and M3dregMerge().
- Deprecation: M3dmapImport()/M3dmapExport() are deprecated.
MbufImport()/MbufExport() can now be used to import/export ply and stl
files.
- Deprecation: M3dmapExtract() is deprecated and replaced by M3dimProject()
and two helper functions M3dimCalibrateDepthMap() and
M3dimCalculateMapSize(). Gap filling can be performed using
M3dimFillGaps(). Partially corrected depth maps and uncorrected depth maps
can be retrieved using M3dmapCopyResult().
- Deprecation: M3dmapArith() is deprecated and replaced with M3dimArith().
In addition to M3dmapArith(), M3dimArith() supports 32-bit unsigned depth
maps.
- Deprecation: M3dmapPut()/M3dmapGet() are deprecated. Accessing the data
from a container can be carried out by MbufPut()/MbufGet().
- Deprecation: M3dmapPtCldCalculate() is deprecated and replaced by
M3dmetDistance().
- Deprecation: M3dmapCopyCond() is deprecated and replaced by M3dimCrop().
- Deprecation: M3dmapSetBox() is deprecated. To define a bounding box,
call M3dimStat(M_BOUNDING_BOX) and then copy the box using
M3dimCopyResult() into a an object allocated with M3dgeoAlloc(M_GEOMETRY).
To define a regular box, use M3dgeoAlloc(M_GEOMETRY), then call
M3dgeoBox().
- Deprecation: M3dmapSetGeometry() is deprecated. To define a geometry, use
M3dgeoAlloc(M_GEOMETRY), followed by M3dgeoBox(), M3dgeoPlane(),
M3dgeoCylinder(), M3dgeoSphere() or M3dgeoLine(). For fitting data to the
geometry, use M3dmetFit().
- Deprecation: M3dmapStat() is deprecated and replaced by M3dimStat() which
works on both depth maps and containers. To compute statistics relative to
another object, use M3dmetStat().
- Deprecation: M_3DMAP_GEOMETRY object is deprecated.
- Deprecation: M_3DMAP_STAT_RESULT is deprecated.
3.04 Code Reader module
If required, declare the #define M_MIL_WARN_ON_DEPRECATED_MCODE
directive set to 0 to discard the compilation warnings associated to
the use of the deprecated Mcode... APIs.
- Deprecation: M_WRITE_SIZE_X/Y and M_WRITE_CELL_NUMBER_X/Y used to inquire
about McodeWrite() requirements given a code model and string are now
deprecated and replaced by new ResultType.
Instead of using:
McodeWrite(ModelId, M_NULL, StringToWrite, M_DEFAULT)
McodeInquire(ModelId, M_WRITE_SIZE_X, &SizeX)
use:
McodeAllocResult(SystemId, M_CODE_WRITE_RESULT, &WriteResultId)
McodeWrite(ModelId, M_NULL, StringToWrite, M_DEFAULT, WriteResultId)
McodeGetResult(WriteResultId, M_WRITE_SIZE_X, &SizeX).
- Deprecation: In McodeAllocResult(), the ControlFlag M_CODE_RESULT is
deprecated and replaced by: M_CODE_GRADE_RESULT, M_CODE_READ_RESULT,
M_CODE_WRITE_RESULT; M_DEFAULT same as M_CODE_GRADE_RESULT.
- Deprecation: In McodeControl() and McodeInquire(), if the code type is
M_DATAMATRIX, the control value M_ECC_n (n different than 200) and
M_ECC_NONE are deprecated.
- Deprecation: In McodeGetResult():
M_NUMBER_PASS replaced by M_PASSED_NUMBER_OF_IMAGES.
M_NUMBER_FAIL replaced by M_FAILED_NUMBER_OF_IMAGES.
M_IMAGE_ID_PASSED replaced by M_PASSED_IMAGES_ID.
M_IMAGE_ID_FAILED replaced by M_FAILED_IMAGES_ID.
M_IMAGE_INDEX_PASSED replaced by M_PASSED_IMAGES_INDEX.
M_IMAGE_INDEX_FAILED replaced by M_FAILED_IMAGE_INDEX.
- Deprecation: In McodeControl(), the control type
M_AIMDPM_CALIBRATION_RESULTS is deprecated and replaced by
M_DPM_CALIBRATION_RESULTS.
- Deprecation: In McodeControl() and McodeInquire(), the control value
M_AIMDPM_GRADING is deprecated and replaced by M_ISO_DPM_GRADING.
- Deprecation: McodeGetResultSingle() has been deprecated and replaced by
the newly changed McodeGetResult(ResultCodeId, ResultIndex, RowOrScanIndex,
ResultType, UserVarArrayPtr)
- Note that the prototype of McodeWrite() has been changed to accommodate
a new Result parameter. While function overloads exist to preserve the
compatibility with former versions, we strongly recommend upgrading to
the new Code Reader API.
- Note that the prototype of McodeDraw() has been changed to accommodate
a new ScanIndex parameter. While function overloads exist to preserve the
compatibility with former versions, we strongly recommend upgrading to
the new Code Reader API.
- Note that the prototype of McodeGetResult() has been changed to
accommodate a new RowScanIndex parameter. While function overloads exist
to preserve the compatibility with former versions, we strongly recommend
upgrading to the new Code Reader API.
- Note to C programmers: function prototypes for McodeGetResult(),
McodeDraw() and McodeWrite() have changed. By default, these functions map
to the old prototype to preserve backward compatibility. C users must add
"#define M_USE_OLD_MCODE_API_IN_C 0" before including "mil.h" to use
the new prototypes.
3.05 Primitives
- MimFilterAdaptive() with the M_NOPEL adaptive filter operation is
deprecated and replaced with M_NOISE_PEAK_REMOVAL, which is faster and
gives better results.
- MimConvolve() with M_SHEN_FILTER and M_DERCIHE_FILTER are deprecated. They
are replaced by the new M_LINEAR_FILTER_IIR_CONTEXT.
Instead of using:
MimConvolve(MilImage, MilDestImage, M_DERICHE_FILTER(M_SMOOTH, 50));
use:
MIL_ID MilLinearFilterIIRContext = MimAlloc(MilSystem,
M_LINEAR_FILTER_IIR_CONTEXT, M_DEFAULT, M_NULL);
MimControl(MilLinearFilterIIRContext, M_FILTER_SMOOTHNESS, 50);
MimConvolve(MilImage, MilDestImage, MilLinearFilterIIRContext);
MimFree(MilLinearFilterIIRContext);
Instead of using:
MimConvolve(PsResultImageId, PsResultImageId,
M_SHEN_FILTER(M_SHARPEN, 50));
use:
MIL_ID MilLinearFilterIIRContext = MimAlloc(MilSystemId,
M_LINEAR_FILTER_IIR_CONTEXT, M_DEFAULT, M_NULL);
MimControl(MilLinearFilterIIRContext, M_FILTER_TYPE, M_SHEN);
MimControl(MilLinearFilterIIRContext, M_FILTER_RESPONSE_TYPE, M_STEP);
MimDifferential(MilLinearFilterIIRContext, PsResultImageId, M_NULL,
M_NULL, M_NULL, PsResultImageId, M_NULL, M_DEFAULT,
M_SHARPEN, M_DEFAULT);
MimFree(MilLinearFilterIIRContext);
Instead of using:
MimConvolve(MilTemplateLumImage, MilGradientLumMaskImage,
M_SHEN_FILTER(M_EDGE_DETECT, GradientSmoothness));
use:
MIL_ID MilLinearFilterIIRContext = MimAlloc(MilSystem,
M_LINEAR_FILTER_IIR_CONTEXT, M_DEFAULT, M_NULL);
MimControl(MilLinearFilterIIRContext, M_FILTER_TYPE, M_SHEN);
MimControl(MilLinearFilterIIRContext, M_FILTER_SMOOTHNESS,
GradientSmoothness);
MimDifferential(MilLinearFilterIIRContext, MilTemplateLumImage, M_NULL,
M_NULL, M_NULL, MilGradientLumMaskImage, M_NULL,
M_DEFAULT, M_GRADIENT, M_DEFAULT);
MimFree(MilLinearFilterIIRContext);
3.06 General
- All Inquires to inquire a string size which end by _SIZE are deprecated.
Now + M_STRING_SIZE must be used instead.
Example: M_CAMERA_MODEL_SIZE ==> M_CAMERA_MODEL + M_STRING_SIZE
- MdispD3D utility DLL is deprecated. MIL users must use the new M3ddisp API.
- The MfuncAllocScript() settings M_INTERPRETER_C_PYTHON34 and
M_INTERPRETER_C_PYTHON35 are deprecated. It is recommended to use
M_INTERPRETER_C_PYTHON37.
4. Fixed bugs
4.01 Classification module
- Fixed bug: MclassControl() with M_TIMEOUT will no longer invalidate the
preprocessing of a context.
- Fixed bug: MclassControl() with M_STOP_PREDICT will no longer prevent further
predicts once applied to a result.
- Fixed bug: MclassPredict() no longer returns erroneous scores (was happening
in rare cases)
4.02 Blob module
- Fixed bug: MblobCalculate() will no longer crash for specific blob shapes,
and when M_MAX_BLOBS is set and M_RETURN_PARTIAL_RESULTS is enabled.
- Fixed bug: MblobGetResult(...+M_AVAILABLE...) can now be used for all
result types even if MblobCalculate() has not been called yet.
- Fixed bug: MblobControl() with M_FERET_AT_PRINCIPAL_AXIS_ANGLE set to
M_ENABLE will no longer calculate all the M_FERETS features.
- Fixed bug: MblobGetResult() will no longer return incorrect values for
world box features when M_WORLD_BOX was not enabled and M_BOX was enabled.
It will report an error instead.
- Fixed bug: MblobGetResult() for calibrated results will no longer take a very
long time (was happening in rare cases).
- Fixed bug: MblobGetResult() will no longer return an incorrect result for
M_MAX_BLOBS_END. This was happening when multi-processing was used and the
blob count reached M_MAX_BLOBS (only for the bottom part of the image).
4.03 Calibration module
- Fixed bug: McalTransformCoordinate() family of functions will no longer
result in a crash when transforming a point on top of the camera pinhole.
This was happening only when using 3D calibration modes.
- Fixed bug: McalTransformCoordinate() functions in M_LINEAR_INTERPOLATION
calibration mode could sometimes be inaccurate or have discontinuities when
the calibration grid was irregular.
- Fixed bug: McalTransformImage() will no longer ignore
M_CORRECT_LENS_DISTORTION_ONLY mode when using a 3D Robotics calibration
context.
- Fixed bug: McalControl() will no longer accept a value of 0 for
M_GRAY_LEVEL_SIZE_Z as documented.
4.04 SureDotOCR® module
- Fixed bug: MdmrRead() will no longer fail to read certain strings because
of a wrong internal image cropping (was happening in rare cases).
- Fixed bug: MdmrRead() will no longer try to allocate an internal child
partially outside the parent buffer. This was happening in rare cases and
was resulting in unexpected behavior.
- Fixed bug: MdmrRead() will be less likely to read a string at the wrong
angle (+- 180 degrees from the expected angle). This was happening in rare
cases when most characters to read were reversible (ex.: 6/9, 8, ... ).
- Fixed bug: MdmrRead() will no longer return a character that does not
respect the font constrained. This was happening in rare cases when using
more than one font.
4.05 Edge Finder module
- Fixed bug: MedgeStream will no longer change the value of
M_RESULT_OUTPUT_UNITS (was happening in rare cases).
4.06 Registration module
- Fixed bug: MregCalculate() will no longer invert the polarity for the
Local Shape image result (M_DRAW_LOCAL_SHAPE_IMAGE) when light vectors'
azimuth were not aligned with the image axis. This was happening in some
cases.
- Fixed bug: MregCalculate() will no longer invert the polarity for the
Mean Curvature image (M_DRAW_MEAN_CURVATURE_IMAGE) when
M_LIGHT_VECTOR_TYPE is set to M_CARTESIAN.
4.07 Pattern Matching module
- Fixed bug: MpatPreprocess() will no longer ignore the "don't care" mask
applied to the model while computing the effective first level when
M_FIRST_LEVEL is set to M_AUTO_CONTENT_BASED.
- Fixed bug: MpatFind() will no longer return an incorrect high score for
some occurrences when searching at angle and when M_SEARCH_ANGLE_ACCURACY
is set to a value other than M_DISABLE or M_DEFAULT (was happening in rare
cases).
- Fixed bug: MpatFind() will no longer return an incorrect position using a
context which contains more than one model, with some having a different
reference offset. This was happening in some cases when M_SEARCH_MODE was
set to M_FIND_ALL_MODELS (or M_DEFAULT) and M_SEARCH_ANGLE_MODE was set to
M_DISABLE.
- Fixed bug: MpatFind() will no longer lead to results not respecting the
M_MIN_SEPARATION_X/Y settings when M_SEARCH_ANGLE_MODE is set to M_DISABLE
and M_FIRST_LEVEL is set to 0.
- Fixed bug: MpatGetResult() with result M_SUM_I, M_SUM_II, M_SUM_IM,
M_SUM_M, M_SUM_MM and M_NUMBER_OF_PIXELS are now available for
M_FIND_BEST_MODELS mode when M_SAVE_SUMS is M_ENABLE.
4.08 Model Finder module
- Fixed bug: MmodDefineFromFile() will no longer misinterpret the scale of
ellipses that are not centered at the origin.
- Fixed bug: MmodDefineFromFile() will no longer crash when an entity of the
CAD DXF file is completely occluded by other entities, which was happening
in rare cases.
- Fixed bug: MmodFind() with an M_GEOMETRIC context will no longer miss
occurrences when a search region is set and when M_FIRST_LEVEL is set to 1
and M_LAST_LEVEL is set to 0.
- Fixed bug: MmodPreprocess() will no longer cause an infinite loop when
used with an M_GEOMETRIC context, which was happening in rare cases.
- Fixed bug : MmodFind() will no longer lead to non repeatable results when
used with M_SHAPE_RECTANGLE context. This was happening in rare cases.
4.09 Metrology module
- Fixed bug: MmetControl(M_REGION_ACCURACY_HIGH) will no longer be ignored
when set to M_DISABLE (was happening in rare cases).
- Fixed bug: MmetCalculate() will no longer lead to wrong measurements when
using a child of a calibrated image as a target image.
- Fixed bug: MmetCalculate() will no longer ignore the angle of the infinite
region for a measured edgel feature.
- Fixed bug: MmetCalculate() will now report an error when a calibration
with an inverted Y axis is used as this is not supported.
- Fixed bug: MmetDraw() will no longer generate a MIL error when drawing an
M_DRAW_TOLERANCE for an M_EDGEL feature and using a graphic context with
M_INPUT_UNITS control sets to M_WORLD.
- Fixed bug: Saving and restoring a context will no longer potentially
change the results of MmetCalculate for M_AREA_CONVEX_HULL and
M_PERIMETER_CONVEX_HULL tolerances.
- Fixed bug: MmetSetRegion() will no longer establish a wrong ring-sector
region when the end angle is set to 360.
- Fixed bug: M_ANGULARITY tolerance computation between two line features
will now take into account the order in which the lines are listed.
- Fixed bug: MmetCalculate() will no longer cause an infinite loop when
calculating an M_CIRCLE feature using M_OUTER_FIT. This was happening in
very rare cases.
4.10 Code Reader module
- Fixed bug: McodeWrite() will no longer generate an invalid code while
generating an M_GS1_DATABAR code type with M_SUB_TYPE set to
M_GS1_DATABAR_EXPANDED or M_GS1_DATABAR_EXPANDED_STACKED.
- Fixed bug: In some situations, McodeRead() will no longer fail to decode
a 144x144 Data Matrix code.
- Fixed bug: McodeGrade() will no longer return certain wrong grades or
results if part of the bar code or its quiet zone falls outside of the
image or given child.
- Fixed bug: McodeRead() will no longer miss code occurrences when decoding
1D bar codes and with M_POSITION_ACCURACY set to M_LOW.
- Fixed bug: McodeWrite() will no longer generate wrong Micro QR version:
M1 version (11x11 modules) or M3 version (13x13 modules). McodeRead() will
no longer incorrectly decode such Micro QR versions.
- Fixed bug: McodeGetResult(...M_CELL_NUMBER_X...) will no longer return the
number of bars and spaces for linear bar code. It will now return the
number of modules, as expected.
- Fixed bug: McodeTrain() will no longer incorrectly return M_SQUARE as a
result for M_DATAMATRIX_SHAPE when the tolerance on a square shape is not
respected.
- Fixed bug: McodeControl() will no longer crash when setting the control
type to M_DPM_CALIBRATION_RESULTS and the control value to a result buffer
Id which is not a result from a McodeGrade() operation. This will now report
an error.
- Fixed bug: Code positions and angle results will now always follow the
same convention when reading a flipped 2D cross row or a 2D matrix code.
The convention previously depended on the code type and/or the value of
M_DECODE_ALGORITHM used.
- Fixed bug: McodeWrite() will no longer return an error if M_CELL_NUMBER_X
is M_ANY or M_DEFAULT and M_CELL_NUMBER_Y is set to a valid value or
vice versa.
- Fixed Bug: McodeWrite() will no longer generate an error when encoding a
valid EAN/UPC string with M_ENCODING set to M_ANY and the string to encode
does not contain any add-on.
- Fixed bug: McodeRead() will no longer return in some situations an
incorrect M_CODE_FLIP or M_ANGLE result after reading a mirrored
rectangular Data Matrix.
- Fixed bug: Calling McodeControl() with M_RESET_FROM_TRAINED_RESULT to fill
a context from an M_CODE_TRAIN_RESULT will no longer crash when the
context Id and the result Id are not allocated on the same system (using
DMIL). Instead, an error will be reported.
- Fixed bug: McodeStream() with operation M_SAVE_REPORT will now write
"GS1 Databar" instead of "RSS-14" or "RSS" for all encoding types
supported with the code type M_GS1_DATABAR.
- Fixed bug: McodeGrade() will no longer return different results when
grading an M_COMPOSITECODE using a read result (SrcResultCodeId) or
directly using the ImageBufId (was happening in rare cases).
- Fixed bug: McodeWrite() will no longer return an error while encoding a
Maxicode with M_ENCODING set to M_ANY or M_DEFAULT. If M_ENCODING is set
to any of those values, a Mode 4 (M_ENC_MODE4) will be encoded.
- Fixed bug: McodeWrite() will no longer return an error while encoding a
Data Matrix with M_ERROR_CORRECTION set to M_ANY or M_DEFAULT. If
M_ERROR_CRRECTION is set to any of those values, an M_ECC_200 will be
encoded.
- Fixed bug: McodeDraw(M_DRAW_BOX) will no longer include the 1D/2D
separator in the 2D box when drawing a composite code with a MicroPDF417
as a 2D part.
- Fixed bug: McodeRead() will no longer fail to read deformed 2D matrix code
if we provide the exact angle of one direction.
- Fixed bug: McodeWrite() will now always generate a barcode image with a
black foreground when the control M_FOREGROUND_VALUE is set to M_ANY.
Before it was dependent on the M_INITIALIZATION_MODE used.
4.11 String Reader module
- Fixed bug: MstrRead() will no longer crash in some rare situations when
using a fontless context and reading characters having a thickness close
to 1.
- Fixed bug: MstrRead() will now update the results when reading using an
ROI totally outside the target image. Before, results from the previous
read were still available.
4.12 Primitives
- Fixed bug: MimBinarizeAdaptive() will no longer crash when
M_THRESHOLD_MODE is set to M_NIBLACK or M_LOCAL_MEAN and the source image
is smaller than M_LOCAL_DIMENSION.
- Fixed bug: MimDistance() will no longer saturate the distance to 32767
when a 16-bit unsigned destination is used; it will saturate the distance
to 65535, as it should.
- Fixed bug: MimMorphic() will no longer saturate according to the source
buffer depth but according to the destination buffer depth, as expected.
The bug was only present when using a structuring element containing
values others than 0 and M_DONT_CARE and using an M_DILATE, M_ERODE,
M_OPEN, M_CLOSE, M_BOTTOM_HAT or M_TOP_HAT operation.
- Fixed bug: MimBinarize() will no longer truncate the condition values when
using a non float buffer.
- Fixed bug: MimConvolve() will no longer overflow when using an IIR filter
context, with M_FILTER_RESPONSE_TYPE set to M_STEP and with an M_SIGNED
destination buffer of the same depth as the source.
- Fixed bug: MimConvolve() with M_SHEN_FILTER and M_DERICHE_FILTER macros
have a bias toward bottom and left. Those macros are deprecated and this
will no longer happen when using the new M_LINEAR_FILTER_IIR_CONTEXT.
- Fixed bug: MimRank() will no longer crash or exhibit an undetermined
behavior when the rank is set to a value larger than the structuring
element number of valid elements. The rank will now saturate to the
structuring element number of valid elements.
- Fixed bug: MimArith() will no longer overflow when the operation is set
to M_EXP_CONST and when using M_FLOAT images (was happening in rare cases).
- Fixed bug: MimConvert()/MbufCopy() BGR24 to YUV16 packed UYVY will no
longer access outside the source buffer. This was happening when the
source image SizeX was a multiple of 32.
- Fixed bug: MimMorphic(...M_THICK/M_THIN...) will no longer prematurely
stop and not perform all the operation iterations when used with a
structuring element having M_OVERSCAN set to M_TRANSPARENT (was happening
in rare cases).
- Fixed bug: M_TRANSPARENT overscan will no longer be incorrectly computed
when the structuring element is not centered for neighborhood operations
(was happening in rare cases).
- Fixed bug: MimMorphic(...M_DILATE/M_ERODE...) with structuring element
filled with 0, with overscan transparent and when using a non centered
structuring element no longer gives an incorrect result on borders, which
was happening in rare cases.
- Fixed bug: MimMorphic(...M_DILATE/M_ERODE...) with structuring element
filled with 0, with overscan transparent will now handle the overscan
properly even when the image is smaller than the structuring element size.
- Fixed bug: MimMorphic() will now handle the overscan properly even if the
structuring element is larger than destination image. The problem was
occurring when M_OVERSCAN was set to M_TRANSPARENT in some situations.
- Fixed bug: MimInquire() will now permit you to retrieve the image mask
information for an M_MATCH_CONTEXT even if the context has not been
preprocessed yet.
4.13 Graphics
- Fixed bug: When creating an arc interactively using MgraInteractive() with
M_ARC_THREE_POINTS, the temporary line between the first two points will
now appear, even when the graphics list is empty.
- Fixed bug: MgraDot() with coordinates in the half-pixel border around the
image will now draw the dot correctly.
Fixed bug: The graphic list event M_GRAPHIC_SELECTION_MODIFIED will no longer
be ignored when selecting multiple graphics at the same time.
- Fixed bug: interactive graphics with M_ROTATABLE and M_RESIZABLE both set
to M_DISABLED will no longer be movable using their invisible bounding box.
- Fixed bug: MgraDraw() will no longer skip some drawing graphics when a
calibration conversion error occurs.
5. Known limitations
5.01 Classification module
- MclassPredict(): images with a type different than 8+M_UNSIGNED do not
properly support regions (MbufSetRegion).
- Applications using MclassTrain() cannot be called in different threads at
the same time if they are in the same process.
- The first call to MclassTrain() can take a long time to start.
5.02 String Reader module
- MstrGetResult(... M_STRING + M_TYPE_TEXT_CHAR, ...) might not work when
using different OS in a DMIL configuration. It leads to a MIL error
message "Invalid Data Type".
5.03 3D display
- Depending on your local version of Windows and the remote version of
Windows, the 3D Displays may not work properly.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 2: Differences between MIL 10 Processing Pack 3 and MIL 10
Processing Pack 2 with with Update 63
Table of Contents for Section 2
1. Overview
2. New functionalities and improvements
2.01 MIL processing specific examples
2.02 Classification module
2.03 Blob module
2.04 Calibration module
2.05 SureDotOCR® module
2.06 Registration module
2.07 Pattern Matching module
2.08 Model Finder module
2.09 Metrology module
2.10 3dMap module
2.11 Code Reader module
2.12 Measurement module
2.13 String Reader module
2.14 Color module
2.15 Primitives
2.16 Graphics
2.17 General
3. Deprecated functionalities
3.01 OCR module
3.02 Registration module
3.03 Code Reader module
3.04 Blob module
3.05 Graphics
4. Fixed bugs
4.01 Blob module
4.02 Calibration module
4.03 SureDotOCR® module
4.04 Pattern Matching module
4.05 Model Finder module
4.06 Metrology module
4.07 3dMap module
4.08 Code Reader module
4.09 Measurement module
4.10 Edge Finder module
4.11 Registration module
4.12 String Reader module
4.13 Primitives
4.14 Graphics
4.15 Utilities
5. Known limitations
5.01 Classification module
5.02 Model Finder module
-------------------------------------------------------------------------------
1. Overview
- MIL 10 Processing Pack 3 includes all the features of
MIL 10 Processing Pack 2 with Update 63.
In addition, MIL 10 Processing Pack 3 includes new processing
functionalities, performance optimizations, and general improvements
such as:
. new: Registration photometric stereo context
. new: Model Finder rectangle shape finder context
. new: Metrology features and tolerances for 3D profile analysis
. new: Code Reader 1D code type detection
. new: Code Reader training from sample images
. new: Classification module
. new: Primitives such as edge preserving filters,
. and many more to discover!
- MIL 10 is the minimum requirement for all upcoming Processing Packs
until the next major release.
- Note: MIL 10 Processing Pack 3 does not support the MIL GPU system for
image processing.
- Windows’ automatic 8.3 file name creation needs to be enabled in order
for the MIL installer to access the temp folder when the user name
contains a space. This option allows Windows to create short file/folder
name aliases for ones with long names for programs, such as the MIL
installer, that don't support spaces in the file/folder names.
Alternatively, the MIL installer needs to run from a user account that
belongs to the administrators group and has no spaces in it. Note that the
same applies for uninstalling MIL.
- The required Visual C++ 2017 Redistributable needs the presence of
KB2919442 and KB2919355. These will need to be obtained and applied before
installing this Processing Pack.
- The MIL Processing Pack requires a Windows installation that supports
device drivers with SHA-2 digital certificates. Consequently, some Windows
7 installations will require that a Windows Monthly Rollup be applied
before the MIL update can be installed.
- Building the MIL examples using Visual Studio 2015 or 2017 also requires
the presence of Windows SDK version 8.1, which is installed from the
Visual Studio setup.
2. New functionalities and improvements
2.01 MIL processing specific examples
- New MIL 10 Processing Pack 3 examples!
- Note that MIL 10 Update 52 is needed to get the images required
for several examples, in particular for the 3D processing examples.
2.02 Classification module
- New: a classification module to predict the category of an input image
using a restored pre-trained convolutional neural network (CNN).
2.03 Blob module
- New: MblobTransform() function to apply operations on an existing blob
result object.
- New: MblobTransform(M_CONTIGUOUS_LABELS) to perform the relabeling of blobs.
- Improvement: MblobFree() has been optimized in speed when calculation
has been perform on large images containing lots of runs.
- Improvement: the convex hull calculation has been accelerated.
- Improvement: MblobGetResult() will no longer return uninitialized values
but report an error for non calculated feret features.
- Improvement: MblobMerge() has been optimized for speed and memory usage.
- Improvement: the M_PRINCIPAL_AXIS_ANGLE numerical stability has been
improved for perfect symmetric blobs.
- New: M_MAX_LABEL_VALUE can now be retrieved using MblobGetResult().
- New: The M_BLOB_IDENTIFICATION_MODE can now be retrieved using
MblobGetResult.
2.04 Calibration module
- New: McalTransformCoordinate3dList() supports point cloud containers and
3D reconstruction contexts for world-to-world conversions.
- New: McalGrid() now supports regions of interest (ROI).
- Improvement: McalList() with M_DISPLACE_CAMERA_COORD or with
M_DISPLACE_RELATIVE_COORD will no longer require setting the principal point.
In 3d mode and after a successful call to McalGrid(M_FULL_CALIBRATION), it
is now no longer required to specify the principal point before a call to
McalList(M_DISPLACE_*).
- Improvement: the detection of chessboard grids has been improved.
2.05 SureDotOCR® module
- New: DOT_FONT_SEMI_5x9.mdmrf font file has been added following the
"SEMI OCR OUTLINES AUX015-1106" specification.
- New: MdmrGetResult() has new results to retrieve the bounding box of a read
string.
- New: MdmrDraw(M_DRAW_STRING_BOX) can draw the string bounding box.
2.06 Registration module
- New: Photometric stereo is now supported to compute the albedo and the
curvature image from a set of images acquired with light from different
directions.
2.07 Pattern Matching module
- Improvement: M_SEARCH_ANGLE_ACCURACY and M_SEARCH_ANGLE_TOLERANCE
can now be set to values smaller than 0.1 degree.
- Improvement: MpatDefine(M_REGULAR_MODEL+M_CIRCULAR_OVERSCAN) now reports a
better error message when the model definition is too close from an image's
border.
- New: M_HAS_DONT_CARE_MASK can now be inquired to retrieve whether a don't
care mask is set for a model.
2.08 Model Finder module
- New: a rectangle shape finder is now supported!
- New: a segment shape finder is now supported!
- Improvement: ellipse finder will now return the closest angle to angular
mid-value when both the angle and its opposite are valid.
- Improvement: MmodFind() with M_SHAPE_CIRCLE and M_SHAPE_ELLIPSE contexts is
now optimized for multi-core architectures.
- New: M_HAS_TARGET_EDGES_SAVED can now be inquired to retrieve if the target
edges have been saved.
- New: M_RESULT_TYPE can now be inquired to retrieve the type of the result
object (i.e. M_GEOMETRIC, M_GEOMETRIC_CONTROLLED, M_SHAPE_CIRCLE,
M_SHAPE_ELLIPSE, or M_SHAPE_RECTANGLE).
2.09 Metrology module
- New: two edgel features can be now compared and their differences can be
extracted.
- Improvement: Radial fit of a segment within a ring sector region has been
improved.
- New: the ordering of the edgels for an edgel feature can now be specified.
- New: the area tolerance between two edgel features can now be calculated.
- New: the edgel features can now be resampled.
- New: copy of edgel feature can now be done for measured and constructed
features.
- New: the area under the curve tolerance is now supported.
- Improvement: the straightness tolerance has been optimized for speed.
- Improvement: the ring sector region definition has been slightly improved.
- New: inquires to retrieve the information about an external edgel feature
added using MmetPut().
- New: the drawing of labels for both measured and constructed edgels is now
supported.
- New: the Max of min distance tolerance between two features is now supported.
- New: the construction of the closest point to a virtual infinite point at
a specific direction is now supported.
- New: M_PERIMETER_SIMPLE tolerance is now supported to compute the
perimeter of a feature.
- New: M_AREA_SIMPLE tolerance is now supported to compute the area of a
feature.
- New: M_PERIMETER_CONVEX_HULL tolerance is now supported to compute
the perimeter of the convex envelope of a feature.
- New: M_AREA_CONVEX_HULL tolerance is now supported to compute the
area of of the convex envelope of a feature.
- Improvement: M_CLOSEST point construction now supports more combinations
of features.
- Improvement: M_MAX_DISTANCE point construction now supports more
combinations of features.
- New: the closest point construction now supports an optional reference
angle constraint.
- New: the distance tolerances now support directional measure by providing
an optional reference angle constraint.
- Improvement: the bisector between two lines can be controlled to construct
the feature according to the largest angular sector, the smallest angular
sector, or the feature ordering.
- New: measured edgels or edgels during a copy operation can now be
post-processed using a median, average, or Gaussian filter.
- Improvement: the furthest point between two arc constructions has been
optimized for both speed and accuracy.
- Improvement: closest/furthest constructed point is now supported from
multiple features.
- Improvement: new feature, M_WARNING status, is now returned when a feature
definition is valid but its construction is not possible. For example a
segment constructed from two points, while the two points appear to be
identical.
- New: the angularity tolerance can now be set between three points.
- New: the length tolerance can now be set for an edgel feature.
- New: parameterized best fit methods are now proposed for best fitted
features.
2.10 3dMap module
- New: M3dmapPut() now supports M_APPEND to append points to an existing
point cloud.
- New: M_ALIGN_COMPLETED status can now be inquired using M3dmapGetResult()
to retrieve whether the M3dmapAlign() operation successfully completed.
- New: M3dmapExport() function to export a point cloud is now supported.
- New: M3dmapSetGeometry() now supports point cloud containers as input
when using M_FIT_POINT_CLOUD.
- New: M3dmapSetGeometry() supports generating and fitting hemispheres.
- New: M3dmapPtCldCalculate() can now calculate the 3d distance between a
point cloud and a geometry (plane or sphere center).
- New: M3dmapCopy() to copy or to move a source point cloud to a
destination point cloud.
- New: M3dmapCopyCond() to copy or to move a source point cloud to a
destination point cloud based on a given condition.
- Improvement: M3dmapArith() and M3dmapStat() now support source depth
maps with different calibrations along the Z axis.
- New: M3dmapStat() now supports M_DEVIATION_MIN, M_Z_MIN and M_Z_MAX.
2.11 Code Reader module
- New: McodeDetect() is now supported to automatically detect codes types in
an image.
- New: McodeTrain() is now supported to automatically optimize context controls
from a set of images.
- New: McodeRead() now supports Character set Extended Channel Interpretations
(ECIs) for M_DATAMATRIX, M_AZTEC, M_QRCODE, M_MAXICODE, M_PDF417,
M_TRUNCATED_PDF417 and M_MICROPDF417.
- Improvement :McodeGrade() will no longer rarely exhibit slight result
differences between grading directly or from a read result for 1d barcode.
- New: McodeGetResult() and McodeGetResultSingle() now supports the result
type M_IS_ECI.
- Improvement: McodeGetResult() will now return more accurate and stable
bounding box corner results for UPC/EAN code types with add-on.
- New: results M_TOP_LEFT_X, M_TOP_LEFT_Y, M_TOP_RIGHT_X, M_TOP_RIGHT_Y,
M_BOTTOM_LEFT_X, M_BOTTOM_LEFT_Y, M_BOTTOM_RIGHT_X, M_BOTTOM_RIGHT_Y,
M_POSITION_X and M_POSITION_Y can now be combined with M_2D_COMPONENT to
retrieve positions for the 2D component of a composite code.
- New: McodeGetResult() and McodeGetResultSingle() now support the
M_TYPE_STRING combination flag.
- Improvement: the threshold mode M_ADAPTIVE is now supported with M_4_STATE
codes.
- Improvement: McodeRead() is more robust reading skewed M_4_STATE codes.
- Improvement: McodeRead() is more robust reading aliased M_4_STATE, M_PLANET
and M_POSTNET codes.
- New: McodeGetResult()/McodeGetResultSingle() now support retrieving the
read results after a call to McodeGrade().
- New: M_GRADE_QUIET_ZONE flag can enable/disable grading the quiet zone for
M_QRCODE, M_MICROQRCODE, and M_DATAMATRIX.
- Improvement: McodeRead() robustness has been improved when
M_CHECK_QUIET_ZONE is set to M_DISABLE.
- Improvement: McodeRead() will now prioritize codes with valid quiet zone,
even when the M_CHECK_QUIET_ZONE is set to M_DISABLE.
- Improvement: McodeRead() is now more robust regarding the dimension of the
spaces for Industrial-25 codes.
- New: M_CELL_NUMBER_X_MIN, M_CELL_NUMBER_X_MAX, M_CELL_NUMBER_Y_MIN,
M_CELL_NUMBER_Y_MAX controls are now available for all 2D codes.
- New: M_MAXICODE, M_PDF417, M_TRUNCATED_PDF417, M_MICROPDF417 now support
the string format M_RAW_DATA.
- Improvement: McodeRead() and McodeWrite() now support Version 18 of the
GS1 General Specifications.
- New: McodeInquire() can now retrieve the list of the supported code types
using M_SUPPORTED_CODE_TYPES.
- Improvement: McodeRead() has been optimized for speed when reading multiple
linear code models.
- Improvement: McodeGrade() has been optimized for speed for DataMatrix codes.
- Improvement: McodeGrade() robustness has been improved for DataMatrix codes.
- New: McodeControl() now supports M_CODE_SEARCH_MODE (M_FAST, M_BEST) to
return either the N first decoded code or the N best codes among all found.
- Improvement: McodeRead() with M_THRESHOLD_MODE set to M_GLOBAL_SEGMENTATION
has been improved for linear code.
- Improvement: McodeRead() has been improved for 1d codes in noisy image when
M_THRESHOLD_MODE is set to M_ADAPTIVE and the M_MINIMUM_CONTRAST is close
or equal to the image contrast.
- Improvement: McodeRead() has been improved for linear codes at angle.
- Improvement: McodeGrade() robustness has been improved for Code 128 codes
as well as for 2D codes.
- Improvement: McodeRead() has been improved to reduce the number of false
positive solutions when reading M_PHARMACODE codes.
- Improvement: The accuracy of the corners of linear codes has been improved
in the presence of distortion.
- Improvement: McodeRead() and McodeGrade() speed have been improved for
QR Code, Micro QR Code or Aztec when M_DECODE_ALGORITHM is set to
M_CODE_DEFORMED.
- Improvement: McodeRead() robustness has been improved for barcode with
poor modulation in blurred images when the M_SCANLINE_HEIGHT is set to a
value different than the default value.
2.12 Measurement module
- New: the M_SUB_REGIONS_NUMBER_USED result can now be retrieved using
MmeasGetResult().
2.13 String Reader module
- Improvement: MstrRead() is now able to read more than one string along
the same line.
2.14 Color module
- New: McolInquire(M_CONTEXT_TYPE) to distinguish between a M_COLOR_MATCHING
context and a M_COLOR_CALIBRATION_RELATIVE context.
2.15 Primitives
- New: MimConvolve() with a M_LINEAR_FILTER_IIR_CONTEXT context to perform
image filtering using infinite impulse response linear filters such as Deriche
or Shen.
- Improvement: MimMorphic(), with M_DILATE/M_ERODE and with structuring
element filled with 0s, has been optimized for speed.
- Improvement: MimDilate() and MimErode() have been optimized further
for speed.
- Improvement: MimHistogramEqualize() has been optimized further for memory
usage with 32-bit sources.
- New: MimArith() now supports M_ABS_SUB, M_AVG, M_MAX_ABS, and M_MIN_ABS
operations.
- New: MimFilterAdaptive() function to perform edge preserving image
filtering techniques such as the bilateral filter.
- New: MbufBayer() now supports a M_ADAPTIVE_FAST demosaicing operation.
- New: MimRemap() function to perform image intensity remapping.
- Improvement: the accuracy of MimConvert() HSL conversions has been improved.
- New: M_REPLICATE overscan mode for the neighborhood operations is
now supported.
- Improvement: MimRotate(), with rotations multiple of 90 degrees, has been
optimized further for speed.
- New: MimConvolve() predefined FIR convolution kernel M_LAPLACIAN_ISO_8
is now supported.
- Improvement: MimTranslate() has been improved in accuracy.
- New: MimTranslate() now supports the M_NEAREST_NEIGHBOR and the M_BICUBIC
interpolation modes.
- New: MimConvert() now supports HSV conversions.
- New: MimArith() now supports the M_SUB_CONST_ABS operation.
- Improvement: Neighborhood operations better support periodic mirror
overscan - that is when the neighborhood operation requires an overscan
dimension larger than the image dimension.
2.16 Graphics
- New: it is now possible to hold the Shift key to restrict the interactive
translation along the X and Y axes.
- New: interactive translation can now be configured using MgraControlList()
with M_MODE_TRANSLATE, M_MODE_TRANSLATE_SHIFT, M_MODE_TRANSLATE_CTRL and
M_MODE_TRANSLATE_ALT.
- Improvement: MgraRect() and MgraRectFill() will no longer create a rectangle
with negative width or height. Rectangle is now defined by its top left
position with positive width and height.
- New: it is now possible to hold the Alt key to interactively rotate graphics
by a multiple of 15 degrees.
- New: interactive rotation can now be configured using MgraControlList()
with M_MODE_ROTATE, M_MODE_ROTATE_SHIFT, M_MODE_ROTATE_CTRL, M_MODE_ROTATE_ALT
and M_ANGLE_SNAPPING_VALUE.
- New: it is now possible to hold the Shift key to keep the same aspect ratio
while resizing graphics.
- New: it is now possible to hold the Ctrl key to interactively resize a
graphic with respect to its center.
- New: it is now possible to hold the Alt key to force a square aspect ratio
while resizing graphics.
- New: interactive rotation can now be configured using MgraControlList() with
M_MODE_RESIZE, M_MODE_RESIZE_SHIFT, M_MODE_RESIZE_CTRL and M_MODE_RESIZE_ALT.
- New: when interactively changing the secondary dimension of a rectangle or
ring, it is now possible to hold the Ctrl key to change it on both sides.
- New: interactive resize can now be configured to be symmetric using
MgraControlList().
- New: interactive secondary dimension control of a graphic
can now be configured using MgraControlList() with the
M_MODE_RESIZE_SECONDARY_DIMENSION_SHIFT/CTRL/ALT controls.
- New: graphics can now be translated interactively using the keyboard.
See MgraControlList() with M_ACTION_... controls for more customization.
- New: graphics can now be rotated interactively using the keyboard.
See MgraControlList() with M_ACTION_... controls for more customization.
- New: graphics can now be resized interactively using the keyboard.
See MgraControlList() with M_ACTION_... controls for more customization.
- New: the corners can now be inquired for most graphics in a graphics list
using MgraInquireList() with M_CORNER_... control flags.
- New: MgraInteractive() now supports M_ROTATE_AROUND_CORNER flag to create
a rectangle with M_POSITION_TYPE set to M_CORNER_AND_DIMENSION.
- New: MgraInteractive() now supports the M_ORIENTED_RECT creation mode to
create a rectangle at 0 degree using 3 clicks only.
- New: MgraInteractive() now supports the M_CIRCLE creation mode to create a
circle using a click-and-drag.
- New: MgraInteractive() now supports the M_AXIS_ALIGNED_RECT creation mode
to create a rectangle at 0 degree using a click-and-drag.
- New: MgraInteractive() now supports the M_AXIS_ALIGNED_ELLIPSE creation
mode to create an ellipse at 0 degree using a click-and-drag.
- New: MgraInteractive() now supports the M_ARC_THREE_POINTS creation mode to
create a circular arc using 3 clicks only.
- New: M_SAME_LOCATION can now be added to M_POSITION_TYPE to change the
reference position of a rectangle in a graphics list without changing its
location in the image.
- New: M_RESIZE_WIDTH and M_RESIZE_HEIGHT can now be used to change a
graphic's dimensions using MgraControlList().
- New: M_MULTIPLE_SELECTION_KEY can now be used to change the combination key
for interactive multiple selection using MgraControlList().
- New: M_EASY_SELECTION can now be enabled using MgraControlList() to allow
selecting graphics interactively using the entire shape, not only the shape
contour.
- New: M_CENTER_X and M_CENTER_Y can now be inquired to retrieve the center
of most graphics in a graphics list using MgraInquireList().
- New: MgraControlList() with M_ACTION_MODIFIER_SPEED can now be used to
specify the combination key to change the speed of interactive edition.
- New: MgraControlList() with M_ACTION_KEY_DELETE can now be used to specify
the key used to delete the selected graphics.
- New: MgraControlList() with M_ACTION_KEY_CANCEL can now be used to specify
the key used to cancel the current interactive graphic edition.
- New: MgraControlList() with M_ACTION_KEYS can now be used to activate
graphic interactive edition with the keyboard.
- Improvement: when interactively creating dots, the vertices are now updated
until the mouse-up event.
- Improvement: when editing a line interactively, the cursor over the end
points is now easier to distinguish from the translation cursor.
- Improvement: the default value for M_SELECTION_RADIUS has been increased
to 8 to ease graphics list interactivity.
- Improvement: interactively resizing a graphic with M_CONSTRAIN_ASPECT_RATIO
set to 1.0 is more consistent with usual resizing behavior.
- New: M_DRAW_DIRECTION specifies the optional arrows to draw for the
graphic primitives.
- New: MgraVectors() and MgraVectorsGrid() to draw vectors in an image
or in a graphics list. MgraVectors() uses provided coordinates and lengths
while MgraVectorsGrid() uses provided x/y vector lengths on pixel grid.
2.17 General
- New: MobjControl()/MobjInquire() with M_OBJECT_USER_DATA_PTR to associate
a user data pointer to a MIL object.
- New: MdigInquire() with M_SOURCE_NUMBER_OF_FRAMES to retrieve the number of
frames from the grab source (avi, folder,...)
- Improvement: MIL error MessageBox will now display all message boxes from
several threads, if any.
- Improvement: Safetypes now support nullptr in C++: new overloads have been
added to support nullptr.
- New: MimRotate() is now supported in MIL Lite for 90 degree rotations.
- New: In a process using both the managed and unmanaged MIL APIs, the unmanaged
MappInquire(M_ERROR) can now return M_THROW_EXCEPTION if this mode has been
set by the managed code.
- New: MsysInquire(M_DISTRIBUTED_MIL_TYPE) to retrieve how the system has been
allocated using Distributed MIL.
- New: MsysInquire(M_DISTRIBUTED_MIL_REMOTE_COMPUTER_NAME) to retrieve the name
of the computer used with MsysAlloc() for a Distributed MIL allocation.
- New: The dongle serial numbers can now be retrieved using
MappInquire(M_KEY_SERIAL_NUMBER) .
- New: MILweb API extension for publishing MIL object data for access from a
browser or another standalone application using the HTML5 WebSocket
communication protocol. Refer to the MIL Releases Notes for MILWeb for more
information.
3. Deprecated functionalities
3.01 OCR module
- MocrControl()/Inquire() M_STRING_SIZE flag is deprecated and replaced
by M_STRING_CHAR_NUMBER.
3.02 Registration module
- MregInquire() M_NUMBER_OF_ELEMENTS flag is deprecated and replaced
by M_NUMBER_OF_REGISTRATION_ELEMENTS.
3.03 Code Reader module
- McodeControl()/McodeInquire() M_DISTORTION flag is deprecated and replaced
by M_DECODE_ALGORITHM.
- Renaming an M_STRING_FORMAT control value from M_GS1_RAW_DATA to M_RAW_DATA.
- M_NUMBER_OF_SCANS_PER_ROW result flag is deprecated and replaced by
M_ROW_NUMBER_OF_SCANS.
- The McodeGetResultSingle()/McodeDraw() macro M_GRADE_RESULT_INDEX(...) is
deprecated and replaced by M_SCAN_RESULT_INDEX(...) and
M_ROW_RESULT_INDEX(...).
- M_GRADE_RESULT_INDEX(ScanReflectanceProfileIndex) macro (McodeGetResultSingle and
McodeDraw) is deprecated and replaced by M_SCAN_RESULT_INDEX(ScanReflectanceProfileIndex)
and M_ROW_RESULT_INDEX(RowIndex).
3.04 Blob module
- M_BLOB_IDENTIFICATION control flag is deprecated and replaced by
M_BLOB_IDENTIFICATION_MODE.
3.05 Graphics
- M_MOVABLE is deprecated and replaced by M_TRANSLATABLE.
- M_FORCE_ASPECT_RATIO_1 is deprecated and replaced by M_SQUARE_ASPECT_RATIO.
- The interactive M_DEFAULT rectangle creation mode is deprecated and
replaced by M_AXIS_ALIGNED_RECT.
- The interactive M_DEFAULT ellipse creation mode is deprecated and replaced
by M_AXIS_ALIGNED_ELLIPSE.
4. Fixed bugs
4.01 Blob module
- Fix: MblobCalculate() will no longer miss recalculating several Feret
features when M_SAVE_RUNS is set to M_DISABLE, M_RETURN_PARTIAL_RESULTS
is set to M_ENABLE, and M_MAX_BLOBS is set to M_ENABLE.
- Fix: MblobGetResult() will no longer potentially run into an infinite
loop when retrieving calibrated angle results with a calibration
containing invalid regions.
- Fix: MblobCalculate(), MblobGetRuns(), MblobDraw(), MblobFill() and
MblobLabel() will no longer potentially crash in rare situation after
an MblobCalculate() with M_RETURN_PARTIAL_RESULTS and M_MAX_BLOBS and
M_TIMEOUT or M_STOP_CALCULATE is enabled.
- Fix: MblobGetResult() will no longer generate an error for general results
such as M_NUMBER when M_RESULT_OUTPUT_UNITS is set to M_WORLD while
the calculated image was not calibrated.
- Fix: MblobSelect() will no longer crash when merging blobs while using a
sorting key.
- Fix: MblobDraw(M_DRAW_HOLES) will no longer crash when holes are not
calculated.
- Fix: Fixed bug : MblobCalculate() will no longer potentially crash when
calculating the M_FERET_AT_PRINCIPAL/SECONDARY_AXIS_ANGLE + M_GRAYSCALE
without having set to calculate the M_FERET_AT_PRINCIPAL/
SECONDARY_AXIS_ANGLE + M_BINARY.
- MblobGetResult() will no longer crash when the M_NUMBER_OF_FERETS is set
to M_INFINITE and M_FERETS is set to M_ENABLE while retrieving other Feret
results not explicitly set to M_ENABLE.
4.02 Calibration module
- Fix: McalGrid() and McalList() with M_LINEAR_INTERPOLATION mode will
no longer fail or result in a crash when using points with very different
spacing in X and Y.
- Fix: McalFree() will no longer potentially result in a crash when an
allocation and a deallocation are performed simultaneously in other
threads.
4.03 SureDotOCR® module
- Fix: MdmrGetResult(M_STRING_NUMBER) will no longer return -1, but instead
will return 0, when no read has been performed.
- Fix: MdmrGetResult(M_STRING_ANGLE) will no longer return an angle value
outside the range [0, 360[.
- Fix: MdmrControlStringModel() will now work properly in C# and VB.net
when using the control type M_ADD_PERMITTED_CHARS_ENTRY with the control
value M_CHAR_LIST.
- Fix: MdmrGetResult(M_FORMATTED_STRING) will no longer return the wrong
result in an ASCII environment.
4.04 Pattern Matching module
- Fix: MpatFind() will no longer potentially crash when the source image is
not M_READ_PREFETCH_ENABLED and the pattern matching model has a SizeX
less than 6 pixels.
- Fix: MpatFind() will no longer generate a MIL Error when the models' search
regions are different and an ROI is associated to the target image. As
documented, the ROI takes precedence.
- Fix: M_PAT_MODEL objects are now tagged as internally allocated objects.
4.05 Model Finder module
- Fix: MmodDraw() will no longer draw invalid annotations when a
calibration context used by MmodFind() has been freed.
- Fix: Drawing CAD models will no longer potentially break some edges.
- Fix: Circle shape finder will no longer return unexpected circle
results when M_FIT_ERROR_WEIGHTING_FACTOR is set to 0.
- Fix: Circle finder will no longer potentially crash, with MP enabled,
under rare circumstances.
- Fix: MmodGetResult(M_ANGLE) will no longer return the angle value in
World units when the M_RESULT_OUTPUT_UNITS is set to M_PIXEL for a
Geometric or a Geometric Controlled context type.
- Fix: MmodFind() with a M_SHAPE_CIRCLE or a M_SHAPE_ELLIPSE context type
will now return the best occurrences when, in some rare situation, the
M_NUMBER control was set to a fix value and that at least the score of one
occurrence was higher than the certainty value.
- Fix: MmodDraw(M_DRAW_POSITION) with an M_SHAPE_CIRCLE or M_SHAPE_ELLIPSE
context type will no longer draw the wrong direction when used with a
calibration and the specified M_RESULT_OUTPUT_UNITS is set to M_PIXEL.
- Fix: MmodDefineFromFile() will no longer fail to open a CAD DXF file
in some rare situation.
4.06 Metrology module
- Fix: The missing MIL.NET overloads for MmetPut() have been added.
- Fix: The region will no longer be displayed when its local frame cannot
be calculated.
- Fix: Tolerance M_VALUE_WARNING_MIN/MAX default values are now equal to
M_EQUAL_PASS_MIN/MAX to fix a misleading default behavior.
- Fix: Constructing a cloned edgel feature will no longer crash for a very
specific derived region while using multi-core acceleration.
- Fix: the intersection between two circles will no longer potentially
lead to the wrong constructed points.
- Fix: the extended intersection between two arcs, or between an arc
and a circle, will no longer potentially lead to the wrong constructed
points.
- Fix: deleting a feature will no longer potentially crash when the
feature has dependencies.
- Fix: the perpendicularity tolerance will no longer lead to potentially
wrong tolerance status when using different minimum and maximum
tolerance limits while the verified feature is not an edge feature.
- Fix: the parallelism tolerance will no longer lead to potentially
wrong tolerance status when using different minimum and maximum
tolerance limits while the verified feature is not an edge feature.
- Fix: the position tolerance of a single feature relative to the default
global frame will no longer be erroneous.
- Fix: the bisector construction will no longer potentially return the
wrong result regarding the order of the line features.
- Fix: feature / tolerance label upper limit is now set to 0x01000000L
to prevent unexpected behaviors.
- Fix: a circle outer fit could potentially lead to the wrong solution.
This is now fixed.
- Fix: the start angle of a constructed arc could potentially be wrongly
set to 360 degrees. This will no longer be allowed.
- Fix: in some rare conditions, associating a reference frame to a feature
using its label value was not possible. This is no longer the case.
- Fix: the angle constraint of a ring sector will no longer be potentially
ignored when calculating a measured feature.
- Fix: the region constraints will no longer be ignored when using
M_COPY_FEATURE_EDGELS with a constructed edgel and when setting the
M_EDGEL_RELATIVE_ANGLE to M_SAME_OR_REVERSE or when setting the
M_EDGEL_ANGLE_RANGE to 360 degrees.
- Fix: Restoring a context will no longer ignore the angularity tolerance
M_ANGLE value, if any.
- Fix: the maximum distance between a circle and an arc will no longer be
potentially erroneous.
- Fix: calculating the bisector between two identical lines will no longer
lead to an invalid construction.
- Fix: Global result status will no longer return pass if there is at least
one warning tolerance status. It will now return a fail status when there
is at least a fail status, or a warning status if there is at least a
warning status and no fail.
- Fix: The ring sector region can now be used to select edgels for the
construction of a segment.
4.07 3dMap module
- Fix: M3dmapImport() will no longer fail loading M_PLY files with
empty 'face' sections.
- Fix: M3dmapImport() will no longer crash when reading an M_STL file
having no triangles.
- Fix: M3dmapImport() will no longer crash when using an existing point
cloud label.
- Fix: M3dmapStat() with M_DEVIATION_MAX will no longer potentially returns
the wrong result for very large depth map image.
4.08 Code Reader module
- Fix: McodeWrite() will no longer return an error when rendering a model,
and another model of the context has an inconsistent input unit.
- Fix: McodeRead() will no longer perform a short read with M_PHARMACODE
codes when the M_SEARCH_ANGLE_STEP is enabled.
- Fix: McodeRead() will no longer fail to read or return a wrong string when
reading an EAN/UPC code specific type of degradation.
- Fix: McodeInquire() will no longer return M_DEFAULT when the inquire type
is M_DOT_SPACING_MIN + M_DEFAULT or M_DOT_SPACING_MAX + M_DEFAULT.
- Fix: McodeRead() will no longer fail decoding 2D code types with a very
small cell size (as small as 1 pixel) when setting proper
M_CELL_SIZE_MIN/MAX values and setting M_USE_PRESEARCH to M_STAT_BASE.
- Fix: McodeRead() will no longer fail reading 1d codes in low contrast
synthetic images when M_THRESHOLD_MODE is set to M_ADAPTIVE.
- Fix: McodeGetResult() and McodeGetResultSingle() will no longer return
wrong values for M_NUMBER_OF_ERRORS and M_NUMBER_OF_ERASURES after reading
or grading an M_DATAMATRIX in rare situation while multi-core acceleration
is used.
- Fix: McodeGetResult() and McodeGetResultSingle() will no longer return
M_STATUS_READ_OK or M_STATUS_GRADE_OK when M_NUMBER and M_TOTAL_NUMBER are
set to M_ALL and no code is found.
- Fix: McodeRead() will no longer potentially erroneously report an error
when decoding a MaxiCode with M_FOREGROUND_VALUE set to M_FOREGROUND_ANY.
- Fix: McodeRead() and McodeGrade() will no longer ignore the control value
M_TOTAL_NUMBER for M_DATAMATRIX codes.
- Fix: McodeGrade() will no longer fail grading 2D matrix codes when the dot
spacing is not null and M_FOREGROUND_VALUE is set to M_FOREGROUND_ANY.
- Fix: McodeRead() will no longer fail reading a Micro QR code version M1
when M_ERROR_CORRECTION is set to M_ECC_L.
- Fix: a code 128 that does not respect the EAN14 constraints will no longer
be read as an EAN14.
- Fix: McodeWrite() will no longer ignore the value of M_CELL_NUMBER_Y when
writing a DataMatrix code.
- Fix: McodeRead() and McodeGrade() will no longer return an error when the
value of M_CELL_NUMBER_X/Y is set to a fixed valid value and
M_CELL_NUMBER_Y/X is set to M_DEFAULT or to M_ANY.
- Fix: McodeRead() and McodeGrade() will now report an error when an invalid
value for M_CELL_NUMBER_X or M_CELL_NUMBER_Y is used while reading an
M_QRCODE, M_MICROQRCODE or M_AZTEC code.
- Fix: McodeRead() will no longer crash in some rare situation while reading
a linear barcode with M_SPEED set to M_VERY_LOW.
- Fix: McodeDraw() will no longer report an error when operations
are accumulated to render multiple features.
- Fix: McodeRead() will no longer fail to read M_GS1_DATABAR stacked code
when the user provides an inaccurate angle.
- Fix: McodeRead() will no longer fail to decode or return a wrong string
for UPC/EAN codes in some rare situation.
- Fix: McodeRead() will no more potentially fail to read 2D matrix codes in
specific situations when the M_MIN_CELL_SIZE is set to a value less than 3.
- Fix: McodeGrade() will no longer potentially crash when reading Code 128
codes, in specific rare situations.
- Fix: McodeInquire() with M_ENCODING will no longer return a wrong value
for M_MICROQRCODE.
- Fix: McodeRead() will no longer cause a heap corruption in very rare
situation while decoding an ANSI X12 encoded DataMatrix.
- Fix: McodeWrite() will no longer potentially encode the wrong string for
M_DATAMATRIX with M_ECC_200.
- Fix: McodeRestore()/McodeStream() will no longer crash when restoring a
context containing an unknown code type.
- Fix: McodeRead() will no longer potentially crash when reading 4-state
codes in images containing no 4-state codes.
- Fix: McodeRead() will no longer potentially fail reading a DataMatrix when
M_DATAMATRIX_SHAPE is set to M_SQUARE.
- Fix: McodeRead() will no longer crash when reading a C or N Australian
encoded 4-state code while the first decoded character is not valid.
- Fix: McodeRead() will no longer potentially return an invalid string when
reading linear codes with M_STRING_SIZE_MAX sets to a non-default value and
M_ERROR_CORRECTION sets to M_ECC_NONE.
- Fix: McodeGetResult() and McodeGetResultSingle() will no longer report an
error when the result type is M_NUMBER_OF_DATA_CODEWORDS + M_2D_COMPONENT
or M_DATA_CODEWORDS + M_2D_COMPONENT.
- Fix: McodeRead() will no longer fail to read 1D codes in some situation
while M_FOREGROUND_VALUE is set to M_FOREGROUND_ANY.
- Fix: McodeFree() will no longer potentially crash when an allocation and
a deallocation are performed simultaneously in different threads.
- Fix: McodeRead() will now decode entirely the EAN/UPC code, including its
add-on, in every situation when M_ENCODING is set to M_ANY.
- Fix: McodeRead() will no longer decode the wrong string for 1D codes
in a very rare situation when M_SEARCH_ANGLE_STEP is set to a specific
value.
- Fix: McodeRead() will no longer return slightly different results
whether the multi-processing is enabled or disabled, in rare situation
for a M_DATAMATRIX code (did not affect the decoded string).
- Fix: McodeGrade(), in some situations for M_DATAMATRIX codes, will no
longer return wrong grade values for the M_EXTENDED_AREA_... grading
result types when the M_EXTENDED_AREA_REFLECTANCE_CHECK is set to M_ENABLE.
- Fix: McodeRead() will no longer fail to read a M_QRCODE or a M_MICROQRCODE
when there in presence of a perspective distortion.
- Fix: McodeGrade(), for a M_QRCODE or a M_MICROQRCODE, will no longer
return a wrong M_GRID_NONUNIFORMITY and M_GRID_NONUNIFORMITY_GRADE in
some cases with perspective distortion.
- Fix: In rare situations, McodeRead() will no longer detect two occurrences
on a same barcode when reading a linear code.
- Fix: McodeRead() will no longer return a wrong angle result when grading
a GS1 Databar code type.
- Fix: For 1D barcode, the returned width will now be more consistent with
the true size of the barcode when the search angle mode is enabled.
- Fix: In rare situation, McodeGrade() will no longer return an invalid
sub-grade of a fixed pattern damage, sub-grade of version and format
information, invalid number of codewords, number of data codewords,
number of error correction codewords, number of error-corrected and
number of erasure when M_GRADING_STANDARD is set to M_AIMDPM_GRADING.
- Fix: The CodeReader module will no longer crash when retrieving the M_STRING
result (McodeGetResult / McodeGetResultSingle) in DMIL configuration when no
code was previously read, graded, trained or detected.
4.09 Measurement module
- Fix: MmeasGetResult() will no longer ignore SecondResultArrayPtr when
retrieving M_DRAW_PROFILE_SCALE_OFFSET.
- Fix: MmeasGetResult() will no longer crash for a M_BOX_EDGEVALUES
result type with the SecondResultArrayPtr set to M_NULL.
4.10 Edge Finder module
- Fix: MedgeGetResult() of the information of a buffer not available in the
result object will no longer lead to a crash.
4.11 Registration module
- Fix: MregCalculate() will no longer take a long time but fail quickly when
a constant color image is provided to a M_STITCHING context type.
4.12 String Reader module
- Fix: MstrSetConstraint() will now validate M_FONT_INDEX(...) properly
when using a font-less context.
- Fix: MstrInquire() will no longer crash when inquiring the M_CONSTRAINT
at a specific index of a remote context allocated using DMIL.
4.13 Primitives
- Fix: MimResize() with M_MAX or M_MIN will no longer fetch data outside the
source image buffer.
- Fix: MimArith() with M_ATAN2 will no longer ignore regions.
- Fix: MimRestore()/MimStream(), applied to an M_MATCH_CONTEXT will no longer
crash when the saved model was not initialized.
- Fix: MimMorphic() with M_MATCH will no longer potentially return the
wrong result for 8xY structural elements on a 32-bit OS.
- Fix: MimGetResult()/MimGetResult1d() will no longer potentially crash when
retrieving M_POSITION_X, M_POSITION_Y, or M_VALUE on an uninitialized
M_EVENT_LIST.
- Fix: MbufGetLine() will no longer return the wrong result or an error when
the line is completely outside of the buffer limits.
- Fix: MbufGetLine() will no longer potentially crash when the line is
horizontal and completely outside of the buffer limits.
- Fix: MbufGetArc() will no longer potentially miss a portion of the arc
when the angle range was 360.
- Fix: MimResize() will no longer potentially crash when using very large
scale values on a 32-bit OS.
- Fix: MimLocatePeak1d() will no longer potentially crash in rare situations
and after a specific sequence of calls
4.14 Graphics
- Fix: MgraArcFill() will no longer potentially write a full row or column
when the radius is less than 0.5.
- Fix: MbufSetRegion() from a graphics list containing high-level module
drawings (MxxxDraw) will no longer crash when a 'rasterizing' option
is used.
- Fix: the angle of a line graphic will no longer be undefined when its end
points are at the same position.
- Fix: rotated rectangles in a graphic list will no longer be incorrectly
rendered when the display has different X/Y zoom factors.
- Fix: moving a vertex of a graphic using MgraControlList() with
M_POSITION_X/Y or M_TRANSLATE_X/Y will no longer miss updating the global
position of the graphic.
- Fix: moving a line end point interactively will no longer forget updating
the line position and angle.
- Fix: interactively translating a polygon or polyline of size 0 will no
longer change a vertex position instead.
- Fix: interactively translating a circle or arc of size 0 will no longer
change its start or end angle instead.
- Fix: interactively resizing a polygon or polyline whose bounding box has
a null width or height will no longer inconsistently resize the graphic.
- Fix: a rectangle with its top left coordinate at (-0.5, -0.5) will no
longer be rendered outside the buffer.
- Fix: MmeasDraw() with M_DRAW_EDGES will now generate a graphic that can be
selected both in pixel units and world units.
- Fix: MgraInteractive() will no longer lead to unexpected behaviors when
mouse events occur outside the MIL display.
- Fix: MdispControl() with M_REGION_OUTSIDE_SHOW or M_REGION_INSIDE_SHOW will
no longer incorrectly display an inherited region of a child buffer.
- Fix: M_POSITION_X and M_POSITION_Y will no longer return invalid values
when interactively defining a zero-sized line using MgraInteractive().
- Fix: interactively editing the start angle or the end angle of an arc will
no longer generate empty arcs.
4.15 Utilities
- Fix: Matrox Profiler will no longer potentially crash when the Windows operating
system decimal symbol preference is set to "," instead of ".".
- Fix: Matrox Profiler will no longer crash when moving a thread while
running under .NET Framework 4.7 runtime.
- Fix: Matrox Profiler will no longer potentially crash when closing it
while "Export to Text File" is under execution.
5. Known limitations
5.01 Classification module
- MclassPredict() called with image other than 8+M_UNSIGNED won't
support regions (MbufSetRegion) properly.
5.02 Model Finder module
- When a circle is found using a M_SHAPE_ELLIPSE context and a search
region is specified, the angle of the returned circle occurrence can be
outside the specified angle search range.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 3: Differences between MIL 10 Processing Pack 2 with Update 63 and
MIL 10 Processing Pack 2
Table of Contents for Section 3
1. Overview
2. New functionalities and improvements
2.1 SureDotOCR®
3. Quick reference
3.1 SureDotOCR®: how to set a string angle
3.2 SureDotOCR®: how to set a string italic angle
3.3 SureDotOCR®: how to delete a char entry
-------------------------------------------------------------------------------
1. Overview
- MIL 10 Update 63 includes new functionalities, performance optimizations
and general improvements for the MIL SureDotOCR® module.
- MIL 10 Update 63 is an update to MIL 10 Processing Pack 2 only.
2. New functionalities and improvements
2.1 SureDotOCR®
- Improvement: the read operation robustness and accuracy has been improved.
- Improvement: the read operation now provides better character localization
stability.
- Improvement: the read operation speed has been greatly improved and
specifying the string angle and orientation can provide additional speed
improvement.
- New: the string italic angle can now be specified using M_ITALIC_ANGLE_MODE
and M_ITALIC_ANGLE control flags.
- New: the string angle or orientation can now be specified using
M_STRING_ANGLE_MODE, M_STRING_ANGLE and M_STRING_ANGLE_INPUT_UNITS control
flags.
- New : MdmrControlStringModel() with M_DELETE_PERMITTED_CHARS_ENTRY is now
supported to delete an entry at a specific index.
- New: MdmrDraw() now supports M_DRAW_MIL_FONT_FORMATTED_STRING.
3. Quick reference
3.1 SureDotOCR®: how to set a string angle
- MdmrControl() with M_STRING_ANGLE_MODE control flag can be set to:
. M_AUTO (M_DEFAULT): the string angle is automatically determined.
. M_ANGLE: a fixed angle specified by the M_STRING_ANGLE control is used to
read a string from left to right with the characters facing upward.
. M_ORIENTATION: a fixed angle specified by the M_STRING_ANGLE control is
used to read a string from left to right with the characters facing
upward or from right to left with the characters facing downward.
- MdmrControl() with M_STRING_ANGLE control flag can be set to a value
from 0 to +360 degrees, or to M_DEFAULT (0 degrees) interpreted according
to the M_STRING_ANGLE_INPUT_UNITS control. The M_STRING_ANGLE control flag
can also be set to M_ACCORDING_TO_REGION to use the angle provided by a
rectangular region associated to the target image.
- MdmrControl() with M_STRING_ANGLE_INPUT_UNITS control flag can be set to:
. M_PIXEL (M_DEFAULT) to use the pixel coordinate system.
. M_WORLD to use the coordinate system of a calibrated target image.
This control is ignored when M_STRING_ANGLE is set to M_ACCORDING_TO_REGION.
3.2 SureDotOCR®: how to set a string italic angle
- MdmrControl() with M_ITALIC_ANGLE_MODE control flag can be set to:
. M_AUTO (M_DEFAULT) : the italic angle is automatically determined.
. M_ANGLE : a fixed italic angle specified by the M_ITALIC_ANGLE
control is used.
Note that the M_ITALIC_ANGLE_MODE must be set to M_AUTO (M_DEFAULT) if
the M_STRING_ANGLE_MODE is set to M_AUTO (M_DEFAULT).
- MdmrControl() with M_ITALIC_ANGLE control flag can be set to a value
between -90 degrees and +90 degrees, or to M_DEFAULT (0 degrees).
The italic angle value must be provided counter clockwise.
Example:
|<-15º>/ /
| / /
| /------/
/ /
/ /
3.3 SureDotOCR®: how to delete a char entry
- MdmrControlStringModel() EntryIndex with M_DELETE_PERMITTED_CHARS_ENTRY can
be set to:
. a value between 0 and M_NUMBER_OF_PERMITTED_CHARS_ENTRIES-1 to delete a
specific entry.
. M_ALL (M_DEFAULT) to delete all entries.
Example :
MdmrControlStringModel(Ctx, M_STRING_LABEL(1), M_POSITION_IN_STRING(2), M_DELETE_PERMITTED_CHARS_ENTRY, 3, M_DEFAULT, M_NULL);
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 4: Differences between MIL 10 Processing Pack 2 and MIL 10
Processing Pack 1
!!! ATTENTION MATROX DESIGN ASSISTANT 4 USERS !!!
The installation of MIL 10, MIL 10 Processing Pack 1, MIL 10 Processing Pack 2
and Matrox Design Assistant 4 may lead to a problematic configuration.
The installation of Matrox Design Assistant 4 Service Pack 1 will
be required to rectify the configuration issue.
Table of Contents for Section 4
1. Overview
2. New functionalities and improvements
2.01 MIL processing specific examples
2.02 Blob module
2.03 Calibration module
2.04 Code Reader module
2.05 Edge Finder module
2.06 Model Finder module
2.07 Registration module
2.08 Metrology module
2.09 Measurement module
2.10 Pattern Matching module
2.11 3dMap module
2.12 String Reader module
2.13 Ocr module
2.14 Dot Matrix module (SureDotOCR®)
2.15 Primitives
2.16 General
2.17 Utilities
2.18 Distributed MIL
2.19 MIL Help
3. Deprecated functionalities
3.01 Calibration module
3.02 Code Reader module
3.03 Pattern Matching module
3.04 Edge Finder module
3.05 Blob module
3.06 Primitives
3.07 General
4. Fixed bugs
4.01 Blob module
4.02 Calibration module
4.03 Code Reader module
4.04 Edge Finder module
4.05 Model Finder module
4.06 Metrology module
4.07 Measurement module
4.08 Pattern Matching module
4.09 Ocr module
4.10 3dMap module
4.11 Color module
4.12 Primitives
4.13 Interactive utilities
4.14 General
5. Known limitations and bugs
5.01 Interactive utilities
5.02 MIL add-on to Microsoft Visual Studio
5.03 MIL Help
5.04 Pattern Matching
5.05 General
-------------------------------------------------------------------------------
1. Overview
- MIL 10 Processing Pack 2 includes all the features of
MIL 10 Processing Pack 1.
In addition, MIL 10 Processing Pack 2 includes new processing
functionalities, performance optimizations, and general improvements
such as:
. new: Calibration support for partial grids,
. new: Calibration support for chessboard grids with fiducial,
. new: Model Finder dedicated ellipse shape matcher,
. new: Registration support for depth from focus,
. new: 3D alignment extended to unorganized clouds of points,
. new: Dot Matrix reader module,
. new: Code Reader default behaviour modes,
. new: Primitives such as the integral image and texture statistics,
. and many more to discover!
- MIL 10 is the minimum requirement for all upcoming Processing Packs
until the next major release.
2. New functionalities and improvements
2.01 MIL processing specific examples
- New MIL MIL 10 Processing Pack 2 examples!
- Note that the MIL 10 Update 37 is needed to get the images required
for several examples, in particular for 3D processing examples.
2.02 Blob module
- New: the blob module now has a new Context based API! See section 3.05 for
more details about the new API.
- New: M_BLOB_INDEX can now be used with MblobGetResult() to refer to blobs
by index instead of label.
- New: M_INDEX_VALUE can now be retrieved using MblobGetResult() to obtain
the index corresponding to a blob label.
- New: M_INDEX_VALUE can now be used as a criterion with MblobSelect().
- New: M_AVAILABLE can now be used with MblobGetResult() to inquire whether
a feature was calculated.
- New: MblobGetResult() with M_TOTAL_NUMBER_OF_CHAINED_PIXELS to retrieve
the total number of chained pixels.
- New: MblobGetResult() with M_TOTAL_NUMBER_OF_CONVEX_HULL_POINTS to retrieve
the total number of convex hull points.
- New: MblobGetResult() with M_TOTAL_NUMBER_OF_FERETS to retrieve the total
number of ferets.
- New: MblobGetResult() with M_NUMBER_OF_FERETS to retrieve the number
of ferets.
- New: MblobGetResult() with M_CALCULATION_TYPE to retrieve whether
MblobCalculate() was done, and with which type of sources (binary only, or
binary and grayscale).
- New: M_BLOB_INDEX can be used with MblobDraw() to refer to blobs by index
instead of label.
- New: with the new Context based API, it is now possible to save, load and
restore a blob context using MblobSave(), MblobRestore() and MblobStream().
- Improvement: MblobSelect() with M_LABEL_VALUE has been accelerated.
- Improvement: MblobCalculate() and MblobGetResult() with sorting keys
associated to enabled features have been accelerated.
2.03 Calibration module
- New: M_GRID_PARTIAL can now be set to M_ENABLE to find calibration grids
that partially fall outside the image.
- New: M_GRID_SHAPE can be set to M_ANY when searching for partial
non-rectangular grids.
- New: M_GRID_HINT_ANGLE_X can now be used to specify the grid orientation
when searching for partial grids.
- New: McalGrid() RowNumber and ColumnNumber parameters can be set to
M_UNKNOWN when searching for partial grids.
- New: McalDraw() now supports M_DRAW_FIDUCIAL_BOX to draw a box around the
grid's fiducial.
- McalGrid() now supports chessboard grids with a Datamatrix fiducial. The
Datamatrix fiducial encodes information about the grid such as the grid
reference point, the orientation as well as the spacing. See McalControl()
with M_GRID_FIDUCIAL.
- New: McalFixture(M_LEARN_OFFSET, M_MODEL_PAT) now supports the new Pattern
Matching API.
- New: McalInquire() M_GRID_UNITS and M_GRID_UNIT_SHORT_NAME can now be
inquired to retrieve the units encoded in the fiducial, after calling
McalGrid() using a chessboard with fiducial.
- New: a chessboard grid sample with fiducial is now provide in the Matrox
Imaging\images directory (ChessboardCalibrationGrid_Fiducial_Letter.pdf).
- Improvement: the accuracy of the M_LINEAR_INTERPOLATION mode has been
improved when the scale difference between the X and the Y coordinates is
large.
- Improvement: McalGrid() is now more robust when using chessboard cells
which contains minor defects.
- Improvement: McalGrid() has been accelerated for chessboard grids.
- Improvement: McalGrid() is now more robust to very dark or very bright
backgrounds.
2.04 Code Reader module
- New: the Code Reader tool now supports the M_INITIALIZATION_MODE to change
the default behaviour of the module alloc-time or control-time. The default
behaviour can be set to a typical reading rate, or to an improved reading
rate. The latter considers by default more degrees of freedom: any
polarity, presence of distortion and non uniformity, angular position,
flipped geometry, etc...
- New: McodeControl() M_STRING_FORMAT can now be combined with M_UTF8 to
retrieve a string using UTF-8 encoding.
- New: The M_FOREGROUND_VALUE can now be set to M_FOREGROUND_ANY for
M_DATAMATRIX, M_MAXICODE, M_PDF417, M_TRUNCATED_PDF417, M_GS1_DATABAR,
M_COMPOSITECODE, M_PLANET, M_POSTNET, M_4_STATE, and M_PHARMACODE.
- New: McodeGrade() now supports ISO 15415 and AIM-DPM grading for Aztec
Rune.
- New: McodeGrade() now supports grading an Aztec code from a read result.
- New: Support the Aztec code detailed segment A and B results for the
Fixed Pattern Damage.
- M_FIXED_PATTERN_DAMAGE_A_GRADE to retrieve grade from segment A.
- M_FIXED_PATTERN_DAMAGE_B_GRADE to retrieve grade from all segments B.
- M_NUMBER_OF_FIXED_PATTERN_DAMAGE_B_SEGMENT to retrieve the number
of segments B.
- M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_A_GRADE to retrieve grade from
segment A in the extended area.
- M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_B_GRADE to retrieve grade from all
segments B in the extended area.
- Improvement: McodeRead() robustness to blurred images and to barcodes with
poor modulation has been improved.
- Improvement: McodeRead(), to read Datamatrix using the
- Improvement: McodeRead(), to read Datamatrix using the
M_PERSPECTIVE_UNEVEN_GRID_STEP mode or the M_UNEVEN_GRID_STEP mode, is now
more robust to aliasing effects.
- Improvement: McodeRead() DataMatrix detection in noisy images has been
improved when M_DISTORTION is set to M_PERSPECTIVE_UNEVEN_GRID_STEP or
M_UNEVEN_GRID_STEP.
- Improvement: McodeRead() DataMatrix detection in noisy images has been
improved when M_DISTORTION is set to M_PERSPECTIVE_UNEVEN_GRID_STEP and
M_THRESHOLD_MODE is set to M_ADAPTIVE.
- Improvement: Support read/write of empty string with 2D codes (except
mode 2 and 3 Maxicodes and Aztec Rune) and cross-row codes. Datamatrix,
QrCode, MicroQrCode, MaxiCode, Aztec, MicroPDF417, truncated PDF417, PDF417
accept 0 as a valid value for M_STRING_SIZE_MIN and M_STRING_SIZE_MAX.
- Improvement: McodeRead() and McodeWrite() now supports the application
identifiers 7005, 7006, 7007, 7008, 7009 and 7010 as specified in
version 15 of the GS1 General Specifications document.
- Improvement: McodeRead() robustness has been improved for M_MICROPDF417
when the M_USE_PRESEARCH is set to M_STAT_BASE.
- Improvement: McodeRead() will now use the angle range to determine the
string to return for M_PLANET and M_POSTNET codes in the rare cases it can
be decoded in both directions.
- McodeGetResult() now accepts + M_STRING_SIZE as combination value to
M_STRING, and M_TEXT to retrieve the size of the corresponding string
including the terminating null character ("\0").
2.05 Edge Finder module
- New: M_CHAIN_APPROXIMATION can now be set to M_WORLD_LINE to perform the
line approximation in world units.
2.06 Model Finder module
- A new specialized M_SHAPE_ELLIPSE finder is now supported.
- New: MmodFind() can now be used with an Edge result target when using
M_SHAPE_CIRCLE context.
- New: MmodInquire() M_HAS_DONT_CARE_MASK, M_HAS_FLAT_REGIONS_MASK,
M_HAS_WEIGHT_REGIONS_MASK inquires to verify if one of the mask has been
defined.
- New: MmodControl() M_COVERAGE_MAX can now be used to set the expected
maximum model coverage for shape (circles and ellipse) finder occurrences.
- New: MmodControl() M_FIT_SCORE_MIN can now be used to set the expected
minimum fit score for shape (circle and ellipse) finder occurrences.
- New: MmodGetResult() M_SCORE_FIT can now be retrieved from a shape finder
(circle and ellipse) result.
- Improvement: MmodFind() in calibrated targets has been improved for
M_SHAPE_CIRCLE context.
- Improvement: the restoration of a context from a file has been accelerated.
- Improvement: MmodFind() has been improved when matching a very large number
of occurrences with an M_GEOMETRIC context.
- Improvement: searching a synthetic circle model with an M_GEOMETRIC context
has been improved for robustness.
- Improvement: MmodFind() with an M_SHAPE_CIRCLE context has been improved
when facing edges shaped in spiral.
- Improvement: greediness of MmodFind() with an M_SHAPE_CIRCLE context has
been dramatically decreased, thus improving the robustness.
- Improvement: model size limitation has been increased up to 4K x 4K.
2.07 Registration module
- New: MregCalculate can now process a stack of images to compute the
corresponding depth from focus index map.
2.08 Metrology module
- New: MmetCalculate() now supports M_NULL as a source image to calculate
contexts which only contains constructed features.
- New: MmetPut() to set external edgels to an M_EXTERNAL_FEATURE constructed
edgel feature.
- New: MmetControl() with M_ACTIVE to enable/disable a feature for the
calculation. The feature and all its dependencies will fail accordingly.
- New: A point feature can now be constructed from a measured point feature.
For example, the construction of a point from the second point of a
measured point feature along a segment region.
- New: A local frame can now be constructed from a segment. The local frame
is positioned at the centre of the segment and is oriented towards the
segment end point.
- New: M_FILL_GAP_ANGLE extraction control flag can now be set for measured
features using MmetControl().
- New: M_FILL_GAP_DISTANCE extraction control flag can now be set for
measured features using MmetControl().
- New: M_CLONE_FEATURE constructed edgel features now support regions. Only
the edgels included in the region are cloned.
- Improvement: edgels are now drawn as crosses when zoomed for better visual
experience.
2.09 Measurement module
- Improvement: the MmeasSetMarker(M_DRAW_PROFILE_SCALE_OFFSET) default
behaviour is now set to M_AUTO_SCALE_PROFILE and M_AUTO_OFFSET_PROFILE.
2.10 Pattern Matching module
- New: the pattern matching module now has a new Context based API!
See section 3.03 for more details about the new API.
- New: MpatGetResult(M_ANGLE) will no longer be restricted to rotated models,
but is available for all model types.
- New: MpatGetResult(M_INDEX) (formerly M_MODEL_INDEX) will no longer be
restricted to M_FIND_BEST_MODELS results, but is available for all results.
- New: In MpatGetResult(), it is now possible to add + M_AVAILABLE to a
ResultType to determine whether a result is available.
- New: MpatControl(M_NUMBER) with the new API, can now be set to 0 for a
specific model in a Context.
- New: MpatStream() with M_MEMORY is supported on remote systems. This was
not the case for the model based deprecated API.
- Improvement: MpatGetResult() can now be cast to a specific data type using
+ M_TYPE_MIL_... combination flags.
- Improvement: MpatFind() with multiple models verifies that all models have
the same size and use the same search region in the target image. If not,
an error is generated.
2.11 3dMap module
- New: M3dmapAlign() no longer has the constraint that the scene point cloud
must be of M_LASER_SCAN type. It now supports M_UNORGANIZED scene point
cloud.
- New: M3dmapAlign now supports the extraction boxes using the following new
control flags M_EXTRACTION_BOX_MODEL and M_EXTRACTION_BOX_SCENE set
to M_USE.
- new: M3dmapAlign() result can now return an M_NO_VALID_POINTS status.
- New: M3dmapControl() M_ERROR_MINIMIZATION_METRIC control flag can be set
to M_POINT_TO_POINT, M_POINT_TO_PLANE+M_USE_MODEL_NORMALS or
M_POINT_TO_PLANE+M_USE_SCENE_NORMALS for an M_PAIRWISE_ALIGNMENT_CONTEXT.
- New: M3dmapImport() now supports M_STL (.stl) 3d CAD file format.
- New: M3dmapGetResult() has new combination constant M_INSIDE_EXTRACTION_BOX
when retrieving the M_NUMBER_OF_3D_POINTS.
- New: M3dmapGet() new M_INSIDE_EXTRACTION_BOX combination flag specifies to
only return the points that are inside the extraction box defined
using M3dmapSetBox().
- New: M3dmapGet() new M_PARTIAL_WRITE combination flag specifies to output
into the destination arrays even if they are not large enough to hold all
the elements.
- New: M3dmapArith() with M_DIST_NN to compute the euclidean distance between
two depth maps.
- New: M3dmapArith() with M_DIST_NN_SIGNED to compute the signed euclidean
distance between two depth maps.
- New: M3dmapArith() with M_VALIDITY_MAP to compute the validity map of the
pixels from two depth maps.
- New: M3dmapInquire() with M_POINT_CLOUD_INDEX_VALUE and
M_POINT_CLOUD_LABEL_VALUE now return M_INVALID when using an invalid point
cloud label or index.
- Improvement: M3dmapArith() has been accelerated for multi-core
architecture (MP).
2.12 String Reader module
- MstrGetResult() now accepts + M_STRING_SIZE as combination value to
M_STRING, M_TEXT, M_FORMATTED_STRING to retrieve the size of the
corresponding string including the terminating null character ("\0").
- MstrInquire() now accepts + M_STRING_SIZE as combination value to
M_CONSTRAINT and M_DEFAULT_CONSTRAINT to retrieve the size of the
corresponding string including the terminating null character ("\0").
2.13 Ocr module
- MocrGetResult() now accepts + M_STRING_SIZE as combination value to
M_STRING to retrieve the size of the corresponding string including the
terminating null character ("\0").
- MocrInquire() now accepts + M_STRING_SIZE as combination value to
M_CONSTRAINT to retrieve the size of the corresponding string including
the terminating null character ("\0").
2.14 Dot Matrix module (SureDotOCR®)
- New: A Dot Matrix Reader module to read a dot matrix string of characters
such as expiry dates and lot numbers. The module supports font definitions,
string model definitions, and string constraints.
2.15 Primitives
- New: MimArith() M_INTEGRAL operation to calculate the integral image of a
source buffer.
- New: MbufInquire() with M_NUMBER_OF_ELEMENT_VALUES_VALID to retrieve the
number of non M_DONT_CARE elements in a structuring element.
- New: MimConvolve() predefined M_SOBEL_X, M_SOBEL_Y, M_PREWITT_X and
M_PREWITT_Y kernels are now supported.
- New: MimProjection() to obtain the sum, minimum, maximum or median
projections of a source buffer.
- New: MimProjection() now accepts an image as destination.
- New: MimStatCalculate() to calculate image statistics, cumulative
statistics, windowed statistics, and co-occurrence based statistics.
- New: MimStatCalculate() can now returns an M_ORIENTATION_DATA_COHERENCE
value associated to the M_ORIENTATION_DATA_MEAN statistics.
- New: MimDraw() with M_DRAW_IMAGE_ORIENTATION now supports a graphic
list as a destination.
- New: MimPolarTransform() can now generate the X/Y LUTs that can be
used with MimWarp(M_WARP_LUT).
- New: MimTranslate() can now be used to generate the equivalent warp matrix.
- New: MimResize() can now be used to generate the equivalent warp matrix.
- New: MimRotate() can now be used to generate the equivalent warp matrix.
- New: MbufGetLine() can now be used to obtain the coordinates of the pixels.
- New: MimWarpList() to warp X, Y arrays of sources positions to X, Y arrays
of destination positions.
- New: MimConvert() can now be used to convert an sRGB image to/from
the CIELab colour space.
- New: MimConvert() can now be used to convert an sRGB image to/from
the CIELch colour space.
- New: McolDistance() now supports the CIE Delta E distance to measure
the distance between colours in the perceptually uniform Lab colour
space.
- New: MimBinarize() auto-threshold modes now supports regions.
- New: MbufChildColor now supports M_ALL_BAND to easily create a full
sized child.
- New: MimDilate() and MimErode() now support M_BINARY_ULTIMATE_ACCUMULATE.
- New: MbufGetList() and MbufPutList() functions to set/get pixel values at
arbitrary positions.
- New: MimProjection() now supports the M_MEAN projection mode.
- New: MbufGetArc() now supports returning the arc positions.
- Improvement: MimResize(M_NEAREST_NEIGHBOR) has been optimized for
binary buffers.
- Improvement: MimPolarTransform(M_POLAR_TO_RECTANGULAR) accuracy has been
improved for large radii.
- Improvement: subpixel annotations are now available for MimLocatePeak1d()
results when calling MimDraw() with M_DRAW_PEAKS + M_CROSS.
- Improvement: MgraArcFill() and MgraArc() results and speed have been
improved.
- Improvement: MbufGetLine() and MbufPutLine() now support providing the
start and end positions outside the buffer limits.
2.16 General
- New: MdigProcess() now supports M_REGION_DELETE_AT_EACH_FRAME to delete
any region that could have been associated to the modified buffer in a
previous iteration.
- New: MfuncAllocScript(M_INTERPRETER_C_PYTHON35) to allocate a script-based
user-defined function for Python 3.5.
- New: MIL for Python now supports the MIL_FONT_NAME() functionality.
- New: MIL for Python now supports the M_PTR_TO_DOUBLE() functionality.
- New: MdigProcess() now supports M_REGION_DELETE_AT_EACH_FRAME to force
deleting any ROI that could have been associated to the modified buffer
in a previous iteration.
2.17 Utilities
- The DispD3D DLL utility now offers to:
- set the initial 3d view of the scene,
- optionally display the point clouds relative coordinate systems,
- optionally display a point cloud's extraction box.
2.18 Distributed MIL
- Improvement: the default cluster type has been changed from single to
multiple clusters, unless Distributed MIL Server Process or Cluster Type
settings in MILConfig have been modified prior to this update.
2.19 MIL Help
- New: The Filters button displays a Filter Values by dialog box with
a list of different possible filters. The Filter Values dialog box can also
be opened using the Filters menu item in the grey bar. Choose the filter you
want, and then press Apply. The filters only apply to values within tables.
Note that filters are not persistent between different pages, or when you
change page display settings, such as the MIL Systems settings or the
detailed/summarized view. A dialog prompt will appear when a setting change
causes your filters to reset.
- New: Function descriptions and parameters can now be collapsed. Click on the arrow
head to the left of the description or parameter title and it will hide (collapse)
everything inside that title. This functionality is similar to clicking on the
arrow heads to the left of a parameter name (in a parameter association table) or
a value (in a table).
- New: Some concepts in the User Guide are now described using animations
(for example, fixturing).
- Improvement: Constant values listed in the index now jump to their description
on their reference page (instead of jumping to the top of the page).
- Improvement: The data type of the address to pass to void parameters is now
always described near the independent parameter value that establishes the
required data type.
3. Deprecated functionalities
3.01 Calibration module
- M_GRID_CORNER_HINT_X is deprecated and replaced by M_GRID_HINT_PIXEL_X.
- M_GRID_CORNER_HINT_Y is deprecated and replaced by M_GRID_HINT_PIXEL_Y.
- McalGrid() combination flags M_Y_AXIS_UP and M_Y_AXIS_DOWN are deprecated
and replaced by M_Y_AXIS_COUNTER_CLOCKWISE and M_Y_AXIS_CLOCKWISE.
- McalInquire() M_Y_AXIS_UP flag is deprecated and replaced by
M_Y_AXIS_DIRECTION.
3.02 Code Reader module
- M_DOT_SPACING is deprecated and replaced by the range of spacing values
defined by M_DOT_SPACING_MIN and M_DOT_SPACING_MAX.
3.03 Pattern Matching module
- The "model based" Pattern Matching API has been deprecated and replaced by
the following "context based" Pattern Matching API:
- MpatAllocAutoModel() is replaced by MpatAlloc() + MpatDefine(M_AUTO_MODEL).
- MpatAllocModel() is replaced by MpatAlloc() + MpatDefine(M_REGULAR_MODEL).
- MpatAllocRotatedModel() is replaced by MpatAlloc() + MpatDefine(M_REGULAR_MODEL)
with the M_SEARCH_ANGLE_MODE set to a specific value, and the
M_SEARCH_ANGLE_DELTA_NEG/POS set to 0.
- MpatCopy() is replaced by MpatDraw().
- MpatFindModel() is replaced by MpatFind().
- MpatFindMultipleModel() is replaced by MpatFind() with
MpatControl(M_SEARCH_MODE).
- MpatPreprocModel() is replaced by MpatPreprocess().
- MpatGetNumber() is replaced by MpatGetResult(M_NUMBER).
- MpatSetAcceptance() is replaced by MpatControl(M_ACCEPTANCE).
- MpatSetAccuracy() is replaced by MpatControl(M_ACCURACY).
- MpatSetAngle() is replaced by MpatControl().
- MpatSetCenter() is replaced by MpatControl() with M_REFERENCE_X and
M_REFERENCE_Y flags.
- MpatSetCertainty() is replaced by MpatControl(M_CERTAINTY).
- MpatSetDontCare() is replaced by MpatMask().
- MpatSetNumber() is replaced by MpatControl(M_NUMBER).
- MpatSetPosition() is replaced by MpatControl() with
M_SEARCH_OFFSET_X, M_SEARCH_OFFSET_Y, M_SEARCH_SIZE_X and
M_SEARCH_SIZE_Y flags.
- MpatSetSearchParameter() is replaced by MpatControl().
- MpatSetSpeed() is replaced by MpatControl(M_SPEED).
If required, declare the #define M_MIL_WARN_ON_DEPRECATED_MPAT
directive set to 0 to discard the compilation warnings associated to
the use of the deprecated Mpat... APIs.
- Note that the prototypes of MpatSave() and MpatRestore() have
been changed for symmetry with other MIL save/restore APIs. While
function overloads exist to preserve the compatibility with former
versions, we strongly recommend upgrading to the new API.
- Note that the prototypes of MpatInquire() and MpatGetResult() have
been changed to accommodate a new Index parameter. While function
overloads exist to preserve the compatibility with former versions,
we strongly recommend upgrading to the new Pattern Matching APIs.
- Note to C users: function prototypes for MpatInquire, MpatGetResult,
MpatSaveA, MpatRestoreA, MpatSaveW, MpatRestoreW, MpatSave, and
MpatRestore have changed. By default, these functions map to the old
prototype to preserve backward compatibility. C users must add
"#define M_USE_OLD_MPAT_API_IN_C 0" before including "mil.h" to use
the new prototypes.
- M_ACCEPTANCE_THRESHOLD is deprecated and replaced by M_ACCEPTANCE
in MpatInquire().
- M_CERTAINTY_THRESHOLD is deprecated and replaced by M_CERTAINTY
in MpatInquire().
- M_CENTER_X is deprecated and replaced by M_REFERENCE_X in MpatInquire().
- M_CENTER_Y is deprecated and replaced by M_REFERENCE_Y in MpatInquire().
- M_POSITION_ACCURACY is deprecated and replaced by M_ACCURACY
in MpatInquire().
- M_NUMBER_OF_OCCURRENCES is deprecated and replaced by M_NUMBER
in MpatInquire().
- M_MIN_SPACING_X is deprecated and replaced by M_MIN_SEPARATION_X
in MpatInquire() and MpatSetSearchParameter().
- M_MIN_SPACING_Y is deprecated and replaced by M_MIN_SEPARATION_Y
in MpatInquire() and MpatSetSearchParameter().
- M_POSITION_START_X is deprecated and replaced by M_SEARCH_OFFSET_X
in MpatInquire().
- M_POSITION_START_Y is deprecated and replaced by M_SEARCH_OFFSET_Y
in MpatInquire().
- M_POSITION_UNCERTAINTY_X is deprecated and replaced by M_SEARCH_SIZE_X
in MpatInquire().
- M_POSITION_UNCERTAINTY_Y is deprecated and replaced by M_SEARCH_SIZE_Y
in MpatInquire().
- M_MODEL_INDEX is deprecated and replaced by M_INDEX in MpatGetResult().
- MpatAllocResult() NbEntries parameter has been replaced by ControlFlag and
must now be set to M_DEFAULT when using the new API.
- MpatInquire(M_ALLOC_TYPE) is deprecated in the new API and replaced by
MpatInquire(M_CONTEXT_TYPE) and MpatInquire(M_MODEL_TYPE).
- Note that an old Pattern Matching model file can be restored into a new
Pattern matching context using MpatRestore() with the ControlFlag set to
M_PAT_MODEL_TO_CONTEXT. To restore additional model files into a context,
use MpatStream(M_LOAD) with the control flag set to M_PAT_MODEL_TO_CONTEXT.
Ex:
MIL_ID CtxId = MpatRestore(MIL_TEXT("Old1.mod"), SysId, M_PAT_MODEL_TO_CONTEXT, M_NULL);
MpatStream(MIL_TEXT("Old2.mod"), M_NULL, M_LOAD, M_FILE, M_DEFAULT, M_PAT_MODEL_TO_CONTEXT, &CtxId, M_NULL);
MpatStream(MIL_TEXT("Old3.mod"), M_NULL, M_LOAD, M_FILE, M_DEFAULT, M_PAT_MODEL_TO_CONTEXT, &CtxId, M_NULL);
MpatStream(MIL_TEXT("Old4.mod"), M_NULL, M_LOAD, M_FILE, M_DEFAULT, M_PAT_MODEL_TO_CONTEXT, &CtxId, M_NULL);
3.04 Edge Finder module
- M_GENERAL_FERET is now deprecated and replaced by M_FERET_GENERAL.
- M_GENERAL_FERET_ANGLE is now deprecated and replaced by
M_FERET_GENERAL_ANGLE.
- M_DRAW_GENERAL_FERET is now deprecated and replaced by
M_DRAW_FERET_GENERAL.
3.05 Blob module
- MblobControl() M_LATTICE is deprecated and replaced by M_CONNECTIVITY.
- M_GENERAL_FERET is deprecated and replaced by M_FERET_GENERAL.
- M_GENERAL_MOMENT is deprecated and replaced by M_MOMENT_GENERAL.
- M_ALL_FERETS is deprecated and replaced by M_FERET_DIAMETERS.
- The "feature list based" blob API has been deprecated and replaced by the
following "context based" blob API:
- MblobAllocFeatureList() is replaced by MblobAlloc().
- MblobCalculate() parameters have changed when using a blob context:
MblobCalculate(IdentImBufId, GrayImBufId, FeatureListId, ResultId)
is now replaced by:
MblobCalculate(ContextId, IdentImBufId, GrayImBufId, ResultId)
- MblobAllocResult() has new parameters for symmetry with other modules.
- MblobGetResult() has a new parameter LabelOrIndex. LabelOrIndex can be
set to M_DEFAULT (= M_GENERAL or M_INCLUDED_BLOBS depending on the
specified ResultType).
- MblobSelectFeature() is replaced by MblobControl() to enable/disable the
calculation of blob features. In addition:
- MblobSelectFeature(M_XXX + M_FERET_CONTACT_POINTS_XXXX) is now:
- MblobControl(M_FERET_CONTACT_POINTS, M_ENABLE)
- MblobControl(M_GROUP_WITH_M_XXX, M_ENABLE)
- MblobSelectFeature(M_XXX + M_SORTn_DOWN/M_SORTn_UP/M_NO_SORT) is now:
- MblobControl(M_SORTn, M_XXX / M_NO_SORT) with
- MblobControl(M_SORTn_DIRECTION, M_SORT_UP / M_SORT_DOWN)
with M_XXX a sortable ResultType.
- MblobSelectFeature(M_AREA) has no MblobControl() since this feature
is always calculated.
- MblobSelectFeature(M_NO_FEATURES) is replaced by
MblobControl(M_ALL_FEATURES, M_DISABLE).
- MblobSelectMoment() is replaced by:
- MblobControl(M_MOMENT_GENERAL + [M_BINARY / M_GRAYSCALE], M_ENABLE)
- MblobControl(M_MOMENT_GENERAL_MODE, M_CENTRAL / M_ORDINARY)
- MblobControl(M_MOMENT_GENERAL_ORDER_X)
- MblobControl(M_MOMENT_GENERAL_ORDER_Y).
- MblobSelectFeret() is replaced by:
- MblobControl(M_FERET_GENERAL) and
- MblobControl(M_FERET_GENERAL_ANGLE)
- MblobGetNumber() is replaced by MblobGetResult(M_NUMBER).
- MblobGetResultSingle() is replaced by MblobGetResult() using the
M_BLOB_LABEL() macro to specify the LabelOrIndex parameter.
- MblobGetRuns() is replaced by MblobGetResult() with M_RUN_X, M_RUN_Y
and M_RUN_LENGTH. The size of the array can be retrieved using
MblobGetResult(M_TOTAL_NUMBER_OF_RUNS).
- MblobFill() is replaced by MblobDraw(M_DRAW_BLOBS).
- The ContextGraId is used to specify the value with which to fill
the blobs.
- Add "M_DRAW_BLOBS_CONTOUR + M_DRAW_HOLES_CONTOUR" to the operation to
specify that the blob's borders should be filled (same result as the
deprecated MblobFill(+M_CONTOUR) operation).
- Note that MblobDraw() does not propagate the calibration.
- The following controls, that previously apply to a blob result, are
deprecated and now apply to the new "context based" blob, so it can be
saved and restored:
- M_BLOB_IDENTIFICATION
- M_FERET_ANGLE_SEARCH_END
- M_FERET_ANGLE_SEARCH_START
- M_FOREGROUND_VALUE
- M_IDENTIFIER_TYPE
- M_CONNECTIVITY (formerly named M_LATTICE)
- M_NUMBER_OF_FERETS
- M_TIMEOUT
- M_PIXEL_ASPECT_RATIO
- M_MAX_BLOBS
- M_RETURN_PARTIAL_RESULTS
- M_SAVE_RUNS
Note to C users: function prototypes for MblobAllocResult and
MblobGetResult have changed. By default, these functions map to the old
prototype to preserve backward compatibility. C users must add
"#define M_USE_OLD_MBLOB_API_IN_C 0" before including "mil.h" to use
the new prototypes.
3.06 Primitives
- MimProject() is deprecated and replaced by MimProjection().
For example, MimProject(SrcId, ProjListOrDestId, M_0_DEGREE) is replaced
by MimProjection(SrcId, ProjListOrDestId, M_0_DEGREE, M_SUM, M_NULL)
- M_DCT8x8 control value is deprecated and replaced by M_DCT8X8.
- MimStat() is deprecated and is replaced by MimStatCalculate().
For example:
MimAllocResult( SystemId, M_DEFAULT, M_STAT_LIST, &StatResultId);
...
MimStat(ImageId, StatResultId, M_SUM + M_MIN, M_IN_RANGE, 3, 24);
MimGetResult(StatResultId, M_SUM, &SumVar);
...
MimFree(StatResultId);
is deprecated and must be replaced by:
MimAlloc( SystemId, M_STATISTICS_CONTEXT, M_DEFAULT, &StatContextId);
MimAllocResult( SystemId, M_DEFAULT, M_STATISTICS_RESULT, &StatResultId);
MimControl(StatContextId, M_SUM, M_ENABLE);
MimControl(StatContextId, M_MIN, M_ENABLE);
MimControl(StatContextId, M_CONDITION, M_IN_RANGE);
MimControl(StatContextId, M_COND_LOW, 3);
MimControl(StatContextId, M_COND_HIGH, 24);
...
MimStatCalculate(StatContextId, ImageId, StatResultId, M_DEFAULT);
MimGetResult(StatResultId, M_SUM, &SumVar);
...
MimFree(StatResultId);
MimFree(StatContextId);
- MimStatMultiple() is deprecated and replaced by MimStatCalculate.
MimAlloc: M_STAT_MULTIPLE_CONTEXT becomes M_STATISTICS_CUMULATIVE_CONTEXT.
MimAllocResult: M_STAT_MULTIPLE_RESULT becomes M_STATISTICS_RESULT.
- MimResize M_FAST and M_REGULAR control values are now deprecated.
3.07 General
- MbufControlNeighborhood() is deprecated and replaced by MbufControl().
- Deprecated MIL functions are not available in the MIL Python Wrapper.
- MappControlObject and MappInquireObject are deprecated and replaced by
MobjControl and MobjInquire.
4. Fixed bugs
4.01 Blob module
- MblobDraw(M_DRAW_CONVEX_HULL) will no longer draw only a single point when
the convex hull contains 2 points.
- MblobCalculate will no longer potentially crash in rare situations
when M_LABELED_TOUCHING is set to M_BLOB_IDENTIFICATION.
- MblobMerge() will no longer crash when the SrcResult2 is set to M_NULL.
- MblobMerge() will no longer crash when using a result invalidated by a
call to MblobControl().
- MblobSelect() with M_BLOB_TOUCHING_IMAGE_BORDERS will no longer lead to
invalid results when using M_MERGE.
- MblobGetResult(M_BLOB_TOUCHING_IMAGE_BORDERS) will no longer lead to
invalid results or potentially crash when some blobs are excluded or
deleted.
- MblobSelect() with M_BLOB_TOUCHING_IMAGE_BORDERS will no longer result in
a crash when M_SAVE_RUNS is set to M_DISABLE.
- MblobDraw() will no longer result in a crash when providing an invalid
label.
- MblobSelect() will no longer crash when using M_MERGE along with sorting
keys.
- MblobSelect() will no longer miss reporting an error when used on a result
which has never been calculated.
- The module will no longer miss reporting an error when M_SAVE_RUNS feature
is disabled wile required for the calculation of the blob convex hull and
for the calculation of the ferets when the number of ferets is set to
infinite.
- MblobControl() with M_FERET_ANGLE_SEARCH_START and/or
M_FERET_ANGLE_SEARCH_END will no longer discard the result when the value
remains unchanged.
- MblobCalculate() will no longer perform a calculation instead of an
expected post-calculation when buffer(s) are allocated on a different
system than the system of the blob objects (context/result).
4.02 Calibration module
- McalSetCoordinateSystem() will behave correctly when using
M_ROTATION_AXIS_ANGLE with a non-unit axis vector.
- McalGrid() will no longer crash when using a circle grid with calibration
dots that consist of pure black pixels only.
- McalFixture(M_LEARN_OFFSET, M_MODEL_MOD) will no longer generate an error
when using a circle shape (M_SHAPE_CIRCLE) context.
- McalGrid() with a non-null GridOffsetZ will no longer lead to invalid
results when transforming coordinates.
- McalGrid() and McalList() will no longer potentially crash due to numerical
instabilities when using the M_3D_ROBOTICS mode.
- McalGrid() will no longer potentially fail calibrating from a square
chessboard oriented at about 45 degrees when no user hint point is
provided.
- McalFixture() with an M_RESULT_PAT will no longer ignore the region angle.
- McalUniform() will no longer forget to reset the relative coordinate
system.
- McalControl() no longer interpret M_DEFAULT as a value for the controls
M_PRINCIPAL_POINT_X and M_PRINCIPAL_POINT_Y.
4.03 Code Reader module
- McodeRead() will no longer rarely fail to read QrCode Model 2 version 1 and
QrCode Model 1 when M_DISTORTION is set to M_UNEVEN_GRID_STEP.
- McodeGrade() will no longer rarely fail to grade QrCode Model 2 version 1
and QrCode Model 1 when M_DISTORTION is set to M_UNEVEN_GRID_STEP.
- McodeRead() will no longer potentially return an M_STATUS_READ_OK and a
wrong position results when a time-out occurs.
- McodeGrade() will no longer return several wrong ISO and AIM-DPM grades
when grading flipped Datamatrix, Aztec, QrCode or MicroQrCode.
- McodeGrade() will no longer potentially return the wrong grade of version
information when grading a QrCode.
- McodeGrade() will no longer return wrong grades for Datamatrix, Aztec,
QrCode or MicroQrCode, when the extended area partly falls outside
the image.
- McodeRead() will no longer crash when reading a composite code using
a small time-out.
- McodeRead() will no longer crash when reading M_MAXICODE with an
M_DOT_SPACING sets to a non null value.
- McodeRead() will no longer return a wrong string when reading a PDF417 or
truncated PDF417 while most of the code error detection capacity is used.
- McodeGrade() will no longer fail when grading a GS1-128 composite code.
- McodeRead() will no longer potentially perform short read of M_PHARMACODE
codes when the M_SEARCH_ANGLE_STEP is enabled.
- The grading of Interleaved 2 of 5 start and end codes will no longer be
wrongly computed.
- M_FOREGROUND_VALUE sets to M_FOREGROUND_ANY will now be correctly handled
for Aztec, MicroQr and Qr codes when M_DOT_SPACING is set to a
non null value.
- McodeGrade() will no longer ignore the M_FLIP control flag value when
grading 1D code.
- McodeGrade() will no longer potentially return a wrong grade for the few
farthest top and bottom scan lines when grading very long 1D barcode
(~450 pixels and more).
- McodeRead() with M_DISTORTION set to M_PERSPECTIVE_UNEVEN_GRID_STEP will
no longer potentially crash in images of high complexity.
- McodeRead() will no longer try decoding the same region twice when
M_USE_PRESEARCH is set to M_STAT_BASE, in some particular situations.
- McodeRead() will no longer potentially fail when reading in a green or blue
child band buffer with an associated region at angle.
- McodeRead() will no longer potentially crash when using a large scan line
height with a position accuracy set to high.
- McodeRead() will no longer convert Aztec code FNC1 to <GS> when the string
format is not set to M_GS1_RAW_DATA.
- McodeRead() will no longer return an invalid string when the number of
detected errors is greater than the error capacity for an Aztec code.
- McodeGrade() for Interleaved 2 of 5 will no longer return an erroneous
Decode Grade 'F' for a given scan reflectance profile.
- McodeRead() could return a wrong position for M_MICROPDF417. This is no
longer the case.
- McodeRead() will no longer return an invalid string for M_4_STATE codes
when the M_ENCODING is set to M_ENC_US_MAIL.
- McodeRead() will no longer ignore the validation of the Delivery Point
Suffix (DPS) at the end of a UK Mail string.
- McodeRead() will no longer ignore the value of the M_CODE_FLIP control for
M_4_STATE, M_PLANET and M_POSTNET codes.
- In very rare situations, McodeRead() could return two results for a single
linear code. This is no longer the case.
- McodeRead() will no longer take into account the value of M_CODE_FLIP
for M_PHARMACODE as it can always be read in both direction. McodeRead()
will only use the search angle range instead.
- McodeRead() and McodeGrade() will no longer terminate with an exception
when reading or grading in some rare situations of QR-Code.
- McodeVerify() potential minor memory leak has been fixed.
- McodeGrade() will no longer potentially return a wrong Datamatrix grades
for L1 and L2 segments for the fixed pattern damage.
4.04 Edge Finder module
- MedgeCalculate will no longer lead to erroneous edges when using a 16 bit
signed source buffer.
- The module will no longer return inconsistent calibrated and uncalibrated
results after applying a rare combination of selections, including a
relational selection.
4.05 Model Finder module
- M_SHAPE_CIRCLE occurrences will no longer be retrieved outside the search
region when using a reference point different from the default centre
point.
- M_SHAPE_CIRCLE context will no longer use an erroneous search region when
the target is calibrated.
- M_SHAPE_CIRCLE context could potentially crash in rare situations.
- Generally speaking, a few bug fixes are improving the robustness of an
M_SHAPE_CIRCLE context.
- MmodFind() will no longer crash when matching an M_GEOMETRIC or an
M_GEOMETRIC_CONTROLLED context with an M_SHAPE_CIRCLE result buffer.
- When using an M_SHAPE_CIRCLE context, the M_SCALE result and the forward
and reverse coefficients will no longer be wrong when returned in pixel
units.
- The M_POLARITY result will no longer be reversed for an M_SHAPE_CIRCLE
context when the Y axis coordinate of the calibration is up.
- MmodControl() with a M_SHAPE_CIRCLE result will no longer generated an
error when the Index is set to M_DEFAULT.
- MmodFind() with an M_SHAPE_CIRCLE context will no longer potentially crash
when the time-out setting is disabled.
- MmodFind() with an M_SHAPE_CIRCLE context will no longer potentially return
an occurrence with a model coverage score greater than 100.
- MmodDraw() will no longer sometime let some initialized pixels when
drawing a model image.
- MmodPreprocess() will no longer potentially crash, in very rare conditions,
for a Geometric Controlled context.
- Two occurrences will no longer potentially share some same edges when
M_SHARED_EDGES is set to M_DISABLE for a M_GEOMETRIC Context.
4.06 Metrology module
- Constructed inner fit and outer fit segment will no longer potentially
result in erroneous result.
- The extremities of a fitted arc will no longer be potentially wrongly
determined when the parent frame is rotated.
- An extremity of a fitted arc will no longer be potentially wrongly found
when its sub-pixel location falls outside the ring sector region.
- Constructed fitted features will no longer ignore the region. The feature
is now determined using the data included into the region.
- The construction of a point derived from circle will no longer lead to
invalid results when moving the parent local frame.
- Cloning a multiple measured point feature will no longer clone a single
point only.
- In rare situations, the fit of a segment could lead to an invalid result.
This is no longer the case.
- MmetStream() will no longer crash when the ObjectIdPtr is set to M_NULL.
- The minimum distance tolerance between a segment and an arc will no longer
return the wrong value, in a specific configuration of the features.
- A constructed intersection point between a circle and another feature, such
as an arc, will no longer potentially lead to the wrong constructed point.
- The minimum distance between a circle and a segment will no longer return
an erroneous result in some configurations of the two geometries.
4.07 Measurement module
- MmeasSetMarker(M_DRAW_PROFILE_SCALE_OFFSET) will no longer report an error
when set to M_AUTO_SCALE_PROFILE or M_AUTO_OFFSET_PROFILE.
4.08 Pattern Matching module
- Calls to MpatFind() (formerly MpatFindModel) will no longer potentially
modify the M_RESULT_OUTPUT_UNITS state.
- MpatPreprocess() (formerly MpatPreprocModel) will now return an error
message when an error occurred during the preprocessing operation and
inquiring M_PREPROCESSED will return M_FALSE in this situation.
- MpatFind() (formerly MpatFindModel) will no longer potentially find less
occurrences than expected when the angular search and the
M_SEARCH_ANGLE_ACCURACY are enabled.
- MpatRestore()/MpatStream() will no longer crash when loading a
corrupted model file.
- Improvement: MpatFind() (formerly MpatFindModel) will no longer use a
different rounding mode to calculate the score when using the AVX2
instruction set.
- In rare situations, MpatFind() (formerly MpatFindModel) could eliminate
good occurrences when using a first level greater than 0. This is no
longer the case.
- In rare situations, MpatFind() (formerly MpatFindModel) could return
occurrences with a null score. This is no longer the case.
- MpatFind() (formerly MpatFindModel) will no longer fail to find
occurrences on the border of the search region.
- MpatFind() (formerly MpatFindModel) will no longer potentially find
occurrences outside the limits of the search region.
- MpatGetResult(M_ANGLE) with M_RESULT_OUTPUT_UNITS result setting set to
M_WORLD was not transformed in world coordinates system. This is no longer
the case.
- The deprecated MpatFindMultipleModel() will no longer return invalid
results when using a region with M_SEARCH_ANGLE set to a value other
than 0. It will now generate an error.
- The deprecated MpatFindMultipleModel(M_FIND_BEST_MODELS) will no longer
wrongly return results relative to the region but relative to the target
image.
- The deprecated MpatSetSearchParameter() with a ControlValue of 1409 will no
longer be erroneously converted to -32768 for ControlType M_ALLOC_OFFSET_X,
M_ALLOC_OFFSET_Y and M_COARSE_SEARCH_ACCEPTANCE.
- The deprecated MpatSetSearchParameter() with a ControlValue of 1409 will no
longer erroneously set the control to its default value for ControlType
M_LAST_LEVEL, M_FAST_FIND, M_MODEL_STEP, M_MIN_SEPARATION_X,
M_MIN_SEPARATION_Y and M_MAX_INITIAL_PEAKS.
- In general, the Pattern Matching parameter checking has been improved with
the new API and is now more strict.
4.09 Ocr module
- MocrGetResult() will no longer crash when retrieving the results with the
M_SELECT_STRING control set to M_ALL.
- MocrGetResult() will no longer pop an error message when retrieving the
result M_STRING + MIL_TEXT_CHAR with the result control M_SELECT_STRING
sets to M_ALL.
4.10 3dMap module
- M3dmapExtract() will no longer generate incorrect associated calibrations
to destination child buffers.
- M3dmapPut() now correctly handles invalid points (at least one of the
coordinate is set to M_INVALID_POINT).
- M3dmapAlign() RMS error M_ALIGN_RMS_ERROR_THRESHOLD and
M_ALIGN_RMS_ERROR_RELATIVE_THRESHOLD stop conditions are now tested
using '<=' as documented instead of '<'.
- M3dmapAlign() will no longer lead to invalid results when using a user
defined pre-alignment while the M_PREALIGNMENT_MODE control is
set to M_CENTROID.
- M3dmapGet() with M_INTENSITY now returns 0 as the intensity when the point
is invalid.
- M3dmapExtract() with a context allocated on a remote system and image
buffers allocated on a local system will now correctly associate a
calibration to the image buffers.
4.11 Color module
- McolTransform() with a relative color calibration context using the
M_COMPUTE_ITEM_STAT method will no longer produce saturated images when
using color triplets as references.
4.12 Primitives
- MimMorphic(), with M_THICK and M_THIN, will no longer ignore values other
than 0, 1 and M_DONT_CARE in a structuring element and will consider
it as ones.
- The M_ORIENTATION_DATA_MEAN statistics will no longer return an invalid
orientation when the whole image is at a multiple of 90 degrees.
- MimResize will now properly generate an error when up-scaling an image
using M_FILL_DESTINATION with the M_AVERAGE, M_MIN and M_MAX modes.
- MimDraw() with M_DRAW_DEAD_PIXELS will now fill the image with 0 and buffer
maximum value instead of zeros and ones.
- MimLocatePeak1d() will no longer miss finding peaks after a peak with its
width exceeding the maximum allowed value was detected.
- MimGetResult1d output type will now correctly defaults to MIL_INT when an
M_FIND_ORIENTATION_LIST was not allocated with the M_FLOAT attribute.
- MimDraw() will no longer erroneously draw an undecimated centered wavelet
decomposition result.
4.13 Interactive utilities
- Flipped DIB buffers were not properly handled in the module UIs. The
copy-paste of a MIL display into a UI will no longer fail.
- The CodeReader GUI will no longer pop an error message when drawing a read
results after reading twice on a disk sequence, using child buffer, and
with the "Do Process All" option enabled.
4.14 General
- The MIL display will no longer ignore the mouse events when the displayed
buffer has a region of interest, the display has an associated graphics
list, and when interactivity is enabled.
5. Known limitations and bugs
5.01 Interactive utilities
- In Inspector, the spacing and spacing variation controls are deprecated
and should have been grayed-out.
5.02 MIL add-on to Microsoft Visual Studio
- If both MIL 32-bits and 64-bits are installed on the same system but only
one of the installations is updated using MIL 10 Processing Pack 1, the
MIL add-on will nevertheless show up in both cases suggesting the most
recent MIL flags.
5.03 MIL Help
- CHM based help files does not work properly under the following
circumstances:
. The operating System is Windows 7 embedded
. The latest Windows updates, including IE11, are applied to the system
. The user is logged in as a standard user (i.e part of the Users group)
A work around is to log in with administrator privileges.
5.04 Pattern Matching
- MpatGetResult() with M_SUM_I, M_SUM_II, M_SUM_IM, M_SUM_M, M_SUM_MM and
M_NUMBER_OF_PIXELS are not available in M_FIND_BEST_MODELS mode although
M_SAVE_SUMS is set to M_ENABLE.
5.05 General
- When using MIL MfuncAllocScript() with Python 2.7.11, you must set the
following additional environment variable PYTHONHOME to your Python
installation path (eg. set PYTHONHOME=c:\python27).
- When using MIL MfuncAllocScript() with Python 3.4.4, you must set the
following additional environment variable PYTHONHOME to your Python
installation path (eg. set PYTHONHOME=c:\python34).
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 5: Differences between MIL 10 Processing Pack 1 and MIL 10
!!! ATTENTION MATROX DESIGN ASSISTANT 4 USERS !!!
The installation of MIL 10, MIL 10 Processing Pack 1 and
Matrox Design Assistant 4 may lead to a problematic configuration.
The installation of Matrox Design Assistant 4 Service Pack 1 will
be required to rectify the configuration issue.
Table of Contents for Section 5
1. Overview
2. New functionalities and improvements
2.01 MIL processing specific examples
2.02 Bead module
2.03 Blob module
2.04 Calibration module
2.05 Code module
2.06 Color module
2.07 Edge Finder module
2.08 Measurement module
2.09 Model Finder module
2.10 Registration module
2.11 String Reader module
2.12 3dmap module
2.13 Regions
2.14 Primitives
2.15 Matrox Profiler
2.16 General
2.17 Interactive utilities
3. Deprecated functionalities
3.01 Measurement module
3.02 Code module
3.03 3dmap module
3.04 Primitives
4. Fixed bugs
4.01 Bead module
4.02 Blob module
4.03 Calibration module
4.04 Code module
4.05 Color module
4.06 Measurement module
4.07 Metrology module
4.08 OCR module
4.09 Registration module
4.10 String Reader module
4.11 3dmap module
4.12 Pattern matching module
4.13 Model Finder module
4.14 Graphic functions
4.15 Primitives
4.16 Interactive utilities
5. Known limitations and bugs
5.01 Code module
5.02 Interactive utilities
5.03 MIL add-on to Microsoft Visual Studio
5.04 MIL Help
-------------------------------------------------------------------------------
1. Overview
- MIL 10 Processing Pack 1 includes all the features of MIL 10.
In addition, MIL 10 Processing Pack 1 includes new processing
functionalities, performance optimizations, and general improvements
such as:
. new support for Aztec and Industrial 2 of 5 codes,
. new color relative calibrations,
. new support and calibration for multiple 3D camera-laser systems,
. new 3D surface alignment operation,
. new extended depth of field operation,
. new dedicated circle shape matcher,
. support for JIT C#, VB and CPython with MIL,
. MIL optimized for AVX2 instructions set,
. new versatile example launcher utility,
. new extensive wavelet transform operations,
. new local segmentation primitive,
. new MIL add-on for Microsoft Visual Studio,
. and many more to discover!
- MIL 10 is the minimum requirement for all upcoming Processing Packs
until the next major release.
2. New functionalities and improvements
2.01 MIL processing specific examples
- New: a new version of the example launcher with categories, previews,
keywords, rich information, and a search utility.
- 30+ new MIL examples!
- Note that the MIL 10 Update 24 is needed to get the images required
for several examples, in particular for 3D processing examples.
2.02 Bead module
- New: M_ADD_FROM_GRAPHIC_LIST to add a bead template defined using a
graphics list.
2.03 Blob module
- New: M_NUMBER_OF_FERETS can be set to M_INFINITE to calculate ferets and
feret-based features with a high accuracy.
- New: M_CONVEX_HULL_PERIMETER is now supported.
- New: M_BLOB_CONTRAST to calculate the difference between the blob min/max
pixel grayscale values is now supported.
- New: M_FERET_ANGLE_SEARCH_START and M_FERET_ANGLE_SEARCH_END to limit the
angular range used to determine the blob's ferets is now supported.
- New: M_FERET_CONTACT_POINTS (ex. M_FERET_MAX_DIAMETER +
M_FERET_CONTACT_POINTS) is now supported.
- New: M_FERET_CONTACT_POINTS_X1,... to retrieve the coordinates of the
corresponding contact points are now supported.
- New: M_MIN_AREA_BOX is now supported.
- New: M_DRAW_MIN_AREA_BOX to draw the M_MIN_AREA_BOX feature is
now supported.
- New: M_MIN_PERIMETER_BOX is now supported.
- New: M_DRAW_MIN_PERIMETER_BOX to draw the M_MIN_PERIMETER_BOX feature is
now supported.
- New: the minimum area enclosing rectangle feature is now available.
2.04 Calibration module
- New: M_CALIBRATION_MODE can now be inquired on any object accepted
by McalInquire.
- New: M_CALIBRATION_CATEGORY to inquire whether the calibration is an
M_2D_CALIBRATION or an M_3D_CALIBRATION.
- New: M_DEPTH_MAP to inquire whether the MIL object is a fully corrected
depth map or not.
- New: M_CALIBRATION_INPUT_DATA to inquire about the type of the data used
to perform the calibration (ex. grid).
- New: M_POINT_CLOUD_CONTAINER_3DMAP McalFixture location type to use the
relative coordinate system of a 3D reconstruction result buffer as a
fixture provider.
- New: M_LASER_3DMAP McalFixture location type to use the laser line
coordinate system of a 3D reconstruction context buffer as a fixture
provider.
- New: McalSetCoordinateSystem can now be called on 3D reconstruction result
buffers to modify the relative coordinate system, which is used to express
points in world coordinates in M3dmapGet and which defines the projection
orientation in M3dmapExtract.
- New: M_LASER_LINE_COORDINATE_SYSTEM is now supported. Its X-axis is
parallel to the laser line and its Y-axis is parallel to the scanning
direction.
- New: McalGetCoordinateSystem can now be called on a 3D reconstruction
result buffer to access the relative coordinate system.
- New: McalGetCoordinateSystem can now be called on a 3D reconstruction
context to access the laser line coordinate system.
- New: McalGrid GridOffsetZ can now be used to specify the thickness of the
grid when performing a 3D calibration.
- Improvement: McalTransformCoordinate/3dList has been optimized for
multi-core (MP) architectures.
2.05 Code module
- New: reading and grading multiple Data Matrix codes is now supported.
- New: M_INDUSTRIAL25 to read Industrial 2 of 5 codes.
- New: M_AZTEC to read and to grade Aztec codes.
- New: Add-On for EAN-8 code is supported.
- New: M_STRING_FORMAT now supports M_GS1_RAW_DATA.
- New: QR codes now support GS1 encoding.
- New: It is now possible to verify a code from the result of the reading
operation using the new McodeGrade function.
- New: M_DRAW_QUIET_ZONE to draw the quiet zone around the decoded code.
- New: M_QUIET_ZONE_INCLUDED result to determine whether the quiet zone is
fully included into the image borders or not.
- New: M_QUIET_ZONE_TOP_... to retrieve the position of the quiet zone.
- New: verification of the extended quiet zone including Asterisk notation.
- New: M_EXTENDED_AREA_REFLECTANCE_CHECK to enable the verification using
the extended area.
- New: M_DRAW_EXTENDED_AREA to draw the extended area when
M_EXTENDED_AREA_REFLECTANCE_CHECK is enabled.
- New: M_EXTENDED_AREA_BOTTOM_... and M_EXTENDED_AREA_TOP_... to retrieve
the position of the extended area.
- New: M_EXTENDED_AREA_QUIET_ZONE_INCLUDED result to determine whether the
extended area and quiet zone are fully included into the image borders
(ISO/IEC 15415:2011).
- New: M_EXTENDED_AREA_CODEWORD_MODULATION to retrieve the modulations for
each codeword in the symbol in the extended area.
- New: M_EXTENDED_AREA_REFLECTANCE_MINIMUM to retrieve the lowest
reflectance (Rmin) of any extended area in a 2D symbol, except Maxicode.
- New: M_EXTENDED_AREA_REFLECTANCE_MAXIMUM to retrieves the highest
reflectance (Rmax) of any extended area in a 2D symbol, except Maxicode.
- New: M_ASTERISK to retrieve if an Asterisk is appended to the overall
symbol grade. Valid for all 2D matrix symbol, except Maxicode.
- New: M_EXTENDED_AREA_SYMBOL_CONTRAST,
M_EXTENDED_AREA_MODULATION_GRADE,
M_EXTENDED_AREA_CODEWORD_MODULATION_GRADE,
M_EXTENDED_AREA_SYMBOL_CONTRAST_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_L1_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_L2_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_QZL1_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_QZL2_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_CLOCKTRACK_SOLID_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_AVERAGE_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_A1_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_A2_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_A3_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_B1_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_B2_GRADE,
M_EXTENDED_AREA_FIXED_PATTERN_DAMAGE_C_GRADE,...
extended area verification results.
- New: M_FIXED_PATTERN_DAMAGE_L1_GRADE, M_FIXED_PATTERN_DAMAGE_L2_GRADE,
M_FIXED_PATTERN_DAMAGE_QZL1_GRADE, M_FIXED_PATTERN_DAMAGE_QZL2_GRADE,...
to retrieve the detailed information about the fixed pattern damages per
segment, for ISO and AIM-DPM verification.
- New: Additional support for the ISO/IEC 15415:2011 verification:
M_REFLECTANCE_MARGIN_GRADE, M_CODEWORD_REFLECTANCE_MARGIN,
M_CODEWORD_REFLECTANCE_MARGIN_GRADE, M_CONTRAST_UNIFORMITY,
and M_CONTRAST_UNIFORMITY_GRADE are now supported.
- New: M_CODEWORD_REFLECTANCE_MARGIN, M_CODEWORD_REFLECTANCE_MARGIN_GRADE
and M_REFLECTANCE_MARGIN_GRADE verification grade are now supported.
- New: M_USE_PRESEARCH now supports M_FINDER_PATTERN_BASE mode for reading
Data Matrix.
- New: M_FINDER_PATTERN_MINIMUM_LENGTH to set the shortest acceptable
length of either "arm" of the finder pattern of a Data Matrix code is
now supported.
- New: M_FINDER_PATTERN_EXHAUSTIVE_SEARCH is now supported for Data Matrix.
- Improvement: improved reading robustness when the M_CELL_NUMBER_X/Y_MIN
or M_CELL_NUMBER_X/Y are specified.
- Improvement: improved robustness when reading a large 2D barcode with
the presearch enabled.
- Improvement: AIM-DPM verification accuracy.
- Improvement: improved robustness when reading Maxi code and BC412 code.
2.06 Color module
- New: a color relative calibration tool which includes
three (3) calibration methods.
- New: McolTransform can apply the color relative calibration
transformation to an image.
2.07 Edge Finder module
- Improvement: the module has been optimized for multi-core (MP)
architectures.
2.08 Measurement module
- New: M_DRAW_PROFILE_SCALE_OFFSET with M_AUTO_OFFSET_PROFILE control value
to specify the offset for the profile drawing operations.
- New: M_DRAW_PROFILE_SCALE_OFFSET with M_AUTO_SCALE_PROFILE control value
to specify the scale for the profile drawing operations.
- Improvement: it is no longer required to specify an edge modifier
(+M_EDGE_FIRST/SECOND) when calling MmeasGetScore with
(M_DISTANCE_FROM_BOX_ORIGIN_SCORE/M_SPACING_SCORE/M_EDGE_INSIDE_SCORE/
M_STRIPE_WIDTH_SCORE + M_RESULT) on a stripe marker.
2.09 Model Finder module
- A new specialized circle finder is now supported.
- Improvement: robustness of the Geometric controlled context
has been improved.
- Improvement: the module has been optimized for multi-core
(MP) architectures.
2.10 Registration module
- New: the module now supports the Extended Depth of Field (EDF) fusion
operation from stacked images.
2.11 String Reader module
- Improvement: the robustness of the fontless context has been improved.
- Improvement: the MstrRead memory exception safety has been improved.
- Improvement: MstrRead baseline deviation calculation has been improved
when reading a string at angle.
2.12 3dmap module
- New: 3D surface-based alignment between a model's point clouds and an
organized target's point cloud is now supported.
- New: M3dmapImport to import (ASCII and binary) PLY 3D CAD files is
now supported.
- New: M3dmapClear to clear the contents of a 3D reconstruction result
buffer is now supported.
- New: M3dmapPut to store world points and their features inside an
unorganized point cloud.
- New: M3dmapGet to retrieve world point coordinates and their features from
one or more point clouds is now supported.
- New: adding controls to filter out the 3D points according to the Z-axis
during laser scanning process.
- New: M_EXTRACTION_RANGE_Z control, used by M3dmapAddScan, to specify the
range of valid values along the Z-axis.
- New: M3dmapCalibrateMultiple to calibrate multiple 3D reconstruction
contexts simultaneously.
- New: M3dmapCalibrate accuracy has been improved.
- New: M_CALIBRATION_STATUS can now return M_GLOBAL_OPTIMIZATION_ERROR when
the convergence of the M3dmapCalibrateMultiple operation fails.
- New: M_CALIBRATED_CAMERA_LINEAR_MOTION+M_CAMERA_LABEL/M_LASER_LABEL(value)
to specify the camera/laser label for a context.
- New: M_CAMERA_LABEL_VALUE and M_LASER_LABEL_VALUE inquires are
now supported.
- New: the extraction of the depth map (M3dmapExtract) now takes into account
the relative coordinate system of the 3D reconstruction result buffer. This
allows merging depth maps from two different result buffers, adjusting the
depth map according to a fitted plane, or creating slices of an object.
- New: M_EXTRACTION_SCALE_MODE can now be set to
M_USE_DESTINATION_CALIBRATION to use the correction information and the
relative coordinate system associated of the destination image buffer.
- New: M_EXTRACTION_OVERLAP to control the behaviour of the depth map
generation when two or more 3D points projects onto the same depth map
pixel.
- New: M_EXTRACTION_SATURATION to set whether or not to saturate Z
coordinates that fall outside the region when generating the depth map.
- New: M_RESULTS_DISPLACEMENT_MODE and M_RESULTS_DISPLACEMENT_Y to set
whether the point coordinates and the extracted depth maps are returned
according to the object's initial position (before scanning) or according
to any other position, including its current position.
- New: M_TOTAL_DISPLACEMENT_Y returns the total conveyor displacement.
- New: M_ASSUMED_PERPENDICULAR_TO_MOTION to inquire whether the "vertical"
laser plane assumption was assumed during calibration.
- New: M3dmapArith() between two depth maps now allows having different
relative coordinate systems.
- Improvement: improved numerical stability when accumulating scans in
continuous mode.
- New: M_LOCATE_PEAK_1D_CONTEXT_ID can now be inquired for a 3D
reconstruction context to retrieve its internal locate peak 1d context.
- New: M_SCAN_LANE_DIRECTION can now be retrieved for a 3D reconstruction
result to obtain the scan lane direction using M3dmapGetResult.
- New: M3dmapSetBox function to easily define the extraction box to use to
generate the depth map using M3dmapExtract.
2.13 Regions
- New: the following primitives now support regions:
. MimArithMultiple
. MimArith
. MimHistogramEqualize
. MimLocateEvent
. MimCountDifference
. MimStat
. MimProject
. MimFindExtreme
. MbufClear
. MbufClearCond
. MimLutMap
2.14 Primitives
- New: MbufBayer now supports an M_NULL destination to determine the white
balancing coefficients only.
- New: MimBinarizeAdaptive to perform a local segmentation operation is
now supported.
- New: MimArith now supports M_ATAN2.
- New: MimHistogramEqualizeAdaptive (CLAHE) is now supported.
- New: MimFindOrientation to determine the dominant orientations in an image
is now supported.
- New: MimStat now supports M_ANGULAR.
- New: MimStat now supports M_ORIENTATION_DATA_MEAN.
- New: MimConvert now supports conversion from RGB to Normalized RGB.
- New: MimHistogram now supports user-defined and automatic binning modes.
- New: MbufClone to easily clone a MIL buffer is now supported.
- New: M_HORIZONTAL_EDGE_SOBEL and M_VERTICAL_EDGE_SOBEL predefined FIR
convolution kernels are now supported.
- New: M_HORIZONTAL_EDGE_PREWITT and M_VERTICAL_EDGE_PREWITT predefined FIR
convolution kernels are now supported.
- New: MimResize now supports M_MAX and M_MIN interpolation modes.
- New: MimWaveletDenoise is now supported.
- New: MimWaveletTramsform to perform wavelet decomposition and
reconstruction is now supported.
- New: MimSetWaveletFilter to specify a user-defined wavelet
is now supported.
- New: the M_FLAT_FIELD_CONTEXT gain constant can be now set to M_AUTOMATIC.
- New: M_EFFECTIVE_GAIN_CONST result to retrieve the value of the
M_FLAT_FIELD_CONTEXT gain constant when automatically determined by
the operation.
- New: MgenLutFunction now supports M_COLORMAP_JET, M_COLORMAP_HOT,
M_COLORMAP_HUE and M_COLORMAP_SPECTRUM to generate pseudo-color palettes.
- Improvement: MgraLine has been specifically optimized for the case of
straight vertical and horizontal lines.
- Improvement: MimLocatePeak accuracy, robustness, and flexibility
have been improved.
- Improvement: MimPolarTransform now supports an angle range
from -360 to 720 degrees.
- Improvement: MimPolarTransform precision has been improved.
2.15 Matrox Profiler
- New: traces into circular memory buffer are now supported.
- New: extensive global statistics information is now supported.
- New: the profiler sessions can now be saved and restored.
- New: the MIL constant are now translated into their equivalent names.
- New: many new interactivity features and improvements.
- More information can be found in the What's New Profiler read-me file!
2.16 General
- Improvement: loading .TIFF/.MIM large images files (>4GB) is now supported.
- New: MbufChildColor2dClip to allocate a clipped child buffer from one or
more specified bands of a previously allocated parent buffer.
- New: MbufChildMove M_CLIP control flag to automatically adjust the
parameters, if necessary, to make sure the child stays within its
parent's buffer limits.
- New: M_OWNER_SYSTEM can now be inquired for the result buffers.
- New: most primitives and modules have been optimized for SSE4.1
and AVX2 instruction sets.
- New: MfuncAllocScript to run MIL code in JIT CPython, C# and VB .NET
environments.
- New: MIL traces can now be performed into circular memory buffers.
- New: MappControl M_TRACE_SAVE_TO_FILE to dump a trace stored in memory
into a file.
- New: MappTrace(M_TRACE_SET_TAG_INFORMATION) to customize the name and the
color associated with a TraceTag is now supported.
- New: MappInquire(M_TRACE_ACTIVE) to retrieve whether traces are enabled.
- New: MthrAlloc M_TRACE_LOG_DISABLE control flag for M_THREAD object type to
disable the log of traces for and at the creation of a thread.
- New: New Distributed MIL protocol "dmilshm" which uses shared memory for
interprocess communication on a same machine (locahost) and which is intended
to replace the use of dmiltcp://localhost.
- Improvement: M...GetResult and M...Inquire combination flag M_TYPE_SHORT
has been renamed M_TYPE_MIL_INT16.
- New: MbufImport(M_MIL+M_WITH_CALIBRATION) can now be used to restore an
image associated to M_DEFAULT_UNIFORM_CALIBRATION if the image has a
constant pixel size and if the image was previously saved using
MbufExport(M_MIL+M_WITH_CALIBRATION). If not, the function will restore
an uncalibrated image.
2.17 Interactive utilities
- MIL add-on to Microsoft Visual Studio 2012 and up to:
. context sensitive help for C++, C# and VB
. statement completion for C++, C#, and VB
. MIL menu and toolbar
- Note that other Microsoft Visual Studio plugins which provide statement
completion may interfere with the MIL extension for Visual Studio.
3. Deprecated functionalities
3.01 Measurement module
- M_POSITION_IN_MARKER is deprecated and replaced by
M_DISTANCE_FROM_BOX_ORIGIN.
- ResultType M_FOUND_BOX_ANGLE is deprecated and replaced by
M_BOX_ANGLE_FOUND.
- ResultType M_TOTAL_SCORE is deprecated and replaced by M_SCORE_TOTAL.
- ResultType M_BOX_EDGE_VALUES_NUMBER is deprecated and replaced by
M_BOX_EDGEVALUES_NUMBER.
- ResultType M_BOX_EDGE_VALUES is deprecated and replaced by
M_BOX_EDGEVALUES.
- M_TOTAL_SCORE is deprecated and replaced by M_SCORE_TOTAL.
- M_FOUND_BOX_ANGLE is deprecated and replaced by M_BOX_ANGLE_FOUND.
- M_POSITION_INSIDE_STRIPE is deprecated and replaced by
M_INCLUSION_POINT_INSIDE_STRIPE.
3.02 Code module
- CodeVerify has been deprecated and replaced by the more flexible
McodeGrade function.
- M_USE_PRESEARCH M_ENABLE control value is deprecated and replaced
by M_STAT_BASE.
- M_MINIMUM_LSIDE_LENGTH control flag is deprecated and replaced
by M_FINDER_PATTERN_MINIMUM_LENGTH.
- M_MAX_GAP control flag is deprecated and replaced
by M_FINDER_PATTERN_MAX_GAP.
3.03 3dmap module
- M_EXTRACTION_SCALE_MODE M_USE_DESTINATION_SCALES control value is
deprecated and replaced by M_USE_DESTINATION_CALIBRATION.
- The M_LASER_DATA result object is deprecated, use M_LASER_CALIBRATION_DATA,
M_DEPTH_CORRECTED_DATA, or M_POINT_CLOUD_CONTAINER instead.
See M3dmapAllocResult documentation.
- M3dmapAddScan with M_RESET, M_CLEAR_DATA, or M_REMOVE_LAST_SCAN is
deprecated and replaced by M3dmapClear() with M_DELETE, M_CLEAR,
and M_REMOVE_LAST_SCAN.
- M3dmapGetResult with M_3D_POINTS_X, M_3D_POINTS_Y, M_3D_POINTS_Z,
M_3D_POINTS_I is deprecated and replaced by the M3dmapGet function.
- M3dmapGetResult with M_DEPTH_CORRECTION_STATE (M_UNCORRECTED,
M_PARTIALLY_CORRECTED or M_FULLY_CORRECTED) is deprecated. 3D result
objects now have specialized types. Use MappInquireObject() with
M_OBJECT_TYPE to inquire the type of object.
- M_PEAK_WIDTH is deprecated. Use M_LOCATE_PEAK_1D_CONTEXT_ID to retrieve
the internally allocated locate peak 1d context and configure it using
MimControl with M_PEAK_WIDTH_NOMINAL and M_PEAK_WIDTH_DELTA.
- M_MIN_INTENSITY is deprecated. Use M_LOCATE_PEAK_1D_CONTEXT_ID to retrieve
the internally allocated locate peak 1d context and configure it using
MimControl with M_MINIMUM_CONTRAST.
- M_ORIENTATION is deprecated. Use M_LOCATE_PEAK_1D_CONTEXT_ID to retrieve
the internally allocated locate peak 1d context and configure it using
MimControl with M_SCAN_LANE_DIRECTION.
3.04 Primitives
- MimLocatePeadk1d now supports a new improved peak detection method. Using
MimLocatePeadk1d without a context object is now deprecated.
- The default overscan for the morphological operations using a custom
structural element, in binary mode, is now M_TRANSPARENT.
MbufControlNeighborhood(StructElement, M_OVERSCAN, M_DISABLE) must be used
to obtain the previous default behaviour.
- M_LAPLACIAN_EDGE predefined FIR convolution kernel is deprecated and
replaced by M_LAPLACIAN_4.
- M_LAPLACIAN_EDGE2 predefined FIR convolution kernel is deprecated and
replaced by M_LAPLACIAN_8.
- M_SHARPEN predefined FIR convolution kernel is deprecated and
replaced by M_SHARPEN_8.
- M_SHARPEN2 predefined FIR convolution kernel is deprecated and
replaced by M_SHARPEN_4.
- M_EDGE_DETECT predefined FIR convolution kernel is deprecated and
replaced by M_EDGE_DETECT_SOBEL_FAST.
- M_EDGE_DETECT2 predefined FIR convolution kernel is deprecated and
replaced by M_EDGE_DETECT_PREWITT_FAST.
- M_HORIZ_EDGE predefined FIR convolution kernel is deprecated.
- M_VERT_EDGE predefined FIR convolution kernel is deprecated.
4. Fixed bugs
4.01 Bead module
- The training operation will no longer crash when providing invalid
world positions.
4.02 Blob module
- After a call to MblobMerge(M_MOVE), the MblobDraw(M_DRAW_BLOBS_CONTOUR)
will no longer draw the contour of the blobs in the first top frame only.
- The MblobMerge operation will no longer result in the blob's convex hulls
being wrongly offset.
- M_CONVEX_HULL_AREA will no longer return an invalid result for a convex
hull made of 2 points.
- Successive calls to MblobCalculate will no longer result in wrong feret
calculations when using multi-core acceleration together with M_WHOLE_IMAGE
or M_LABELED_TOUCHING modes.
- M_CONVEX_HULL_AREA, M_CONVEX_HULL_COG_X and M_CONVEX_HULL_COG_Y could be
wrongly computed under specific circumstances. This is no longer the case.
- BlobInquire will no longer crash when inquiring another flag
than M_OWNER_SYSTEM.
- MblobReconstruct will no longer crash when providing a NULL seed buffer.
It will now report an error.
4.03 Calibration module
- McalDraw(...+ M_DRAW_..._COORDINATE_SYSTEM) will no longer lead to an
infinite loop in the presence of degenerated calibrations.
- McalFixture with M_RESULT_PAT will no longer lead to the wrong result when
the M_RESULT_OUTPUT_UNITS of the pattern matching result object is manually
set to M_WORLD.
- McalRestore() will no longer restore a calibration object with a wrong
relative coordinate system when restoring from a file containing an image
with a constant pixel size and a relative coordinate system that is not
coplanar to the absolute coordinate system. It will report an error.
Use MbufImport(M_MIL+M_WITH_CALIBRATION).
- M_DISPLACE_CAMERA_COORD will no longer ignore the M_CALIBRATION_PLANE.
4.04 Code module
- McodeRead will no longer crash when reading in an image smaller than 3
pixels with M_USE_PRESEARCH set to M_ENABLE.
- McodeDraw index for M_DRAW_REFLECTANCE_PROFILE will no longer be
wrongly validated.
- When reading a 1D code, the M_STATUS will no longer be M_STATUS_READ_OK
but M_STATUS_NO_RESULT_AVAILABLE when no code has been found.
- McodeRead will no longer crash when reading Postnet/Planet code at angle
in a very large buffer.
- Reading a Data Matrix code will no longer fail when the top right corner of
the code falls outside the image, for uneven grid and perspective
uneven grid modes.
- Data Matrix, QrCode, and MicroQrCode decode algorithm has been fixed when
number of error correction was odd. Small 10x10 and 12x12 Data Matrix will
no longer be erroneously found within larger codes or textured areas.
- McodeRead will no longer fail reading multiple identical overlapped 1D
codes.
4.05 Color module
- McolDistance will no longer miss propagating the calibration of the source
to the destination.
- McolProject will no longer miss propagating the calibration of the source
to the destination.
- Restoring a color context with McolStream(... M_RESTORE ...) in DMIL will
no longer lead to a new allocated MIL_ID with an invalid object type.
4.06 Measurement module
- MmeasGetResult/Single(M_EDGEVALUE_PEAK_WIDTH / M_EDGE_WIDTH) will now
return both edge widths of a stripe marker.
- MeasGetResult/Single will no longer crash when trying to retrieve some
result types not supported by a measurement result.
- MmeasDraw(M_DRAW_LINE) on a measurement result will no longer crash if no
result was calculated.
- MmeasSaveMarker was not returning immediately after parameter checking
detect an error. This is no longer the case.
- MmeasSetMarker(M_BOX_ANGLE_REFERENCE) was not validating the value provided
by the user. This is no longer the case.
- MmeasSetMarker(M_BOX_ANGLE_MODE) was not validating the value provided by
the user. This is no longer the case.
- MmeasSetMarker(M_BOX_ANGLE_TOLERANCE) was not validating the value provided
by the user. This is no longer the case.
- MmeasSetMarker(M_SEARCH_REGION_ANGLE_INTERPOLATION_MODE) was not validating
the value provided by the user. This is no longer the case.
- Under a specific set of conditions, M_EDGE_INSIDE_SCORE will no longer
prevent finding a marker.
- When a score-offset is used for a given feature, the feature's score was
not dropping to zero when the feature was outside the range. It now does.
- MmeasGetResult(M_BOX_EDGEVALUES) will no longer crash when the number of
submarkers change from "more than one" to "a single one" between successive
calls to MmeasFindMarkers.
- MmeasSetMarker(M_BOX_SIZE) will no longer accept a null value. The control
must be set to a value greater than 0.
- The mean angle (for either edges or stripes) returned the M_MEAN
combination flag is used will no longer be erroneous.
- MIL_TYPE_MIL_ID was not supported for inquiring about MIL_ID's information
such as the M_OWNER_SYSTEM. It is now supported.
- Calculation of the edge limits has been improved for very small smoothed
edges.
- MmeasGetResult(M_STRIPE, M_SPACING) will no longer require edge modifiers
to get the spacing from its first/second edges while it did not when not
there is no calibration involved.
- MmeasSetMarker will no longer update the first value when the first value
is valid but not the second.
4.07 Metrology module
- Drawing an edgel feature will no longer draw all the edgels, but the
active ones.
- Metrology will no longer wrongly support the orientation of 3D calibration
systems.
- The calibrated drawing of arc regions will no longer lead to erroneous
annotations when using low accuracy annotations.
- The parametric angle of a constructed local frame will no longer wrongly
follow the global frame, but its local reference frame.
- The coverage measure of the fitted features will no longer be wrongly
calculated in the presence of image aspect ratio.
4.08 OCR module
- A very rare situation of MocrReadString crash has been fixed.
- MocrCalibrateFont will no longer fail when changing the ROI or the
step value.
4.09 Registration module
- MregSetLocation called with parameter type M_COPY_REG_RESULT after a call
with M_WARP_4_CORNER, M_WARP_4_CORNER_REVERSE, or M_WARP_POLYNOMIAL, will
no longer result in a memory leak.
4.10 String Reader module
- String Reader will no longer change the global SPI_SETFONTSMOOTHING system
parameter when defining the system font.
- M_STRING_ANGLE_DELTA_NEG will no longer control the positive delta value
and M_STRING_ANGLE_DELTA_POS will no longer control the negative
delta value.
- The preprocess step will no longer crash in the presence of large
characters in the font images.
- A potential out of memory error has been fixed when reading a string using
a fontless context.
- MstrRead will no longer potentially dilate the user image buffer when using
regions at 0 degrees and when the control M_THICKEN_CHAR is used.
- MstrInquire will no longer crash when a NULL destination is provided.
4.11 3dmap module
- Filling gaps with M_MIN or M_MAX will no longer result in different
decisions in the depth map versus the intensity map.
- M3dmapGetResult() with M_NUMBER_OF_3D_POINTS on an empty result buffer
will no longer return an invalid index error, but 0.
4.12 Pattern matching module
- MpatAllocModel call will no longer perform unnecessary computations when
M_AUTO_CONTENT_BASED is not used.
- MpatAllocModel will no longer crash when one of the provided offset is
negative.
- In rare situations, MpatFindModel could lead to unstable results when using
a search region and the fast find mode.
4.13 Model Finder module
- MmodDraw will now give priority to the destination buffer calibration
when drawing the edges of a context.
4.14 Graphic functions
- MgraRectAngle will no longer ignore the rotation of the relative coordinate
system when drawing the shape at 0 degrees in world units with the
conversion mode set to M_PRESERVE_SHAPE_AVERAGE.
- When calling MbufSetRegion(M_RASTERIZE) with a graphics list that has its
interactivity enabled, the interactive handles will no longer wrongly
appear in the rasterized region.
4.15 Primitives
- MimMorphic(M_DILATE, M_ERODE, M_THICK and M_THIN), in binary mode and with
the overscan set to replace min or max, will no longer use the wrong
overscan value. This affected only board systems that may perform
the operation.
- MimResize(M_BILINEAR) accuracy has been fixed for very large buffers.
- MimDraw will no longer modify the unchanged pixels of the destination
when drawing a stat-multiple result.
- MimDraw will no longer crash when drawing the model image or the mask image
in a large destination buffer.
- MimLocateEvent will no longer return an invalid MIL_INT result when
the result ID is M_NULL.
- MimCountDifference will no longer return an invalid MIL_INT result when
the result ID is M_NULL.
- MimEdgeDetect will no longer potentially crash when the source image is
a child band of a YUV buffer.
- MimPolarTransform(M_POLAR_TO_RECTANGULAR) accuracy of the returned
destination sizes have been fixed.
- MimRank, in binary mode and with signed destination buffer, will no longer
return incorrect results when the rank values of the structuring element
are greater than 128 for an 8-bits destination buffer, or greater than
32768 when the destination buffer is 16-bits.
- MimResize will no longer access data outside the source buffer for rare
specific conditions and specific scale values.
4.16 Interactive utilities
- A GDI memory leak has been fixed for the Model Finder, Edge Finder,
and String Reader GUIs.
- A stability issue has been fixed when using the column ordering in
the result table or the module's UIs.
- The new Model Finder circle matcher has been added to the Model Finder UI.
5. Known limitations and bugs
5.01 Code module
- Grading on read result is not yet supported for newly added Aztec codes.
- Grading is not supported for the special 11×11 Aztec code "rune".
5.02 Interactive utilities
- In Inspector, the spacing and spacing variation controls are deprecated
and should have been grayed-out.
5.03 MIL add-on to Microsoft Visual Studio
- If both MIL 32-bits and 64-bits are installed on the same system but only
one of the installation is updated using MIL 10 Processing Pack 1, the
MIL add-on will nevertheless show up in both cases suggesting the most
recent MIL flags.
5.04 MIL Help
- CHM based help files does not work properly under the following
circumstances:
. The operating System is Windows 7 embedded
. The latest Windows updates, including IE11, are applied to the system
. The user is logged in as a standard user (i.e part of the Users group)
A work around is to log in with administrator privileges.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 6: Differences between MIL 10 and MIL 9 Processing Pack 2 with
Update 56
Table of Contents for Section 6
1. Overview
2. New functionalities and improvements
2.01 MIL processing specific examples
2.02 Bead module
2.03 Blob module
2.04 Calibration module
2.05 Code module
2.06 Color module
2.07 Edge Finder module
2.08 Measurement module
2.09 Metrology module
2.10 Model Finder module
2.11 OCR module
2.12 Pattern matching module
2.13 Registration module
2.14 String Reader module
2.15 3dmap module
2.16 Graphic functions
2.17 Regions
2.18 Primitives
2.19 Matrox Profiler
3. Deprecated functionalities
3.01 General
3.02 Bead module
3.03 Calibration module
3.04 Color module
3.05 Measurement module
3.06 Metrology module
3.07 Model Finder
3.08 Edge Finder
3.09 Interactive utilities
4. Fixed bugs
4.01 General
4.02 Bead module
4.03 Blob module
4.04 Calibration module
4.05 Code module
4.06 Color module
4.07 Edge Finder module
4.08 Measurement module
4.09 Metrology module
4.10 Model Finder module
4.11 OCR module
4.12 Pattern matching module
4.13 Registration module
4.14 String Reader module
4.15 3dmap module
4.16 Graphic functions
4.17 Regions
4.18 Primitives
4.19 Interactive utilities
5. Known limitations
5.01 Graphic functions
5.02 Interactive utilities
5.03 Regions
5.04 Metrology
5.05 Matrox Profiler
5.06 Streaming functions
-------------------------------------------------------------------------------
1. Overview
- MIL 10 includes all the features of MIL 9 Processing Pack 2.
In addition, MIL 10 includes new processing functionalities, performance
optimizations and general improvements.
- MIL 10 is the minimum requirement for all upcoming Processing Packs
until the next major release.
2. New functionalities and improvements
2.01 MIL processing specific examples
- The /Processing/ directory includes new MIL processing specific examples
(see MIL Control Center -> Directories -> MIL examples): a multiprocessing
framework, various product inspections, such as bottle cap and label
verification, color analysis for food inspection of product
identification, 3D object localization, etc., as well as additional 3D
camera interfacing examples.
- The MIL Examples have been updated and use the graphics list object
to display the result with zoomable subpixel annotations.
- A new MIL example has been added which illustrates the benchmarking of
processing operations.
2.02 Bead module
- New: Support to train and to validate stripe-type bead templates with
non-uniform widths.
- New: M_BEAD_EDGE bead template type, to train and to validate the presence
and position of an edge-type bead along a path.
- New: M_NUMBER_FOUND result type to retrieve the number of bead sections
found for a given bead template.
- New: M_TRAINING_PATH with M_CIRCLE and associated M_TEMPLATE_CIRCLE_...
control types to parametrically define a circular bead template.
- New: M_TRAINING_PATH with M_SEGMENT and associated M_TEMPLATE_SEGMENT_...
control types to parametrically define a line segment bead template.
- New: M_INDIVIDUAL_WIDTH_NOMINAL to retrieve the user defined bead section
width for a given bead template.
- New: M_TRAINED_SIZE, M_TRAINED_POSITION_X, M_TRAINED_POSITION_Y, and
M_TRAINED_INDIVIDUAL_WIDTH_NOMINAL to retrieve the trained information
for a given bead template.
- New: M_INTENSITY_NOMINAL_MODE control type to select whether to
train and to validate the color of a stripe-type bead.
- New: M_INTENSITY_NOMINAL control type to define the expected nominal
color of a given bead template.
- New: M_INTENSITY_DELTA_POS and M_INTENSITY_DELTA_NEG control
types to set color tolerance of a given bead template.
- New: MbeadDraw M_FAIL_INTENSITY_COLOR_MIN and M_FAIL_INTENSITY_COLOR_MAX
to display the failed bead sections regarding the color range criterion.
- New: M_ANGLE_ACCURACY_MAX_DEVIATION control to help improve the accuracy
of the width measure when the bead section deviates from the expected
path.
- New: MbeadTemplate M_SET_WIDTH_NOMINAL to set a specific width value for
each user point in order to define bead templates with non-uniform width.
- New: MbeadControl M_WIDTH_NOMINAL_MODE can be set to M_AUTO_UNIFORM or
M_AUTO_CONTINUOUS to either train a bead template with a uniform or with a
non uniform width.
- New: MbeadGetResult M_CLOSURE to retrieve whether a template is closed.
- New: MbeadGetResult M_OFFSET_MAX to retrieve the offset value of the
maximum offsetted bead section of a template.
- New: MbeadGetResult M_OFFSET_MAX_INDEX to retrieve the index of the
maximum offsetted bead section of the template.
- New: MbeadGetResult M_STATUS_FOUND to retrieve the status to determine if
at least one bead section has not been found for a template.
- New: MbeadGetResult M_STATUS_OFFSET to retrieve the status to determine if
at least one bead section offset is out of the tolerance for a template.
- New: MbeadGetResult M_STATUS_WIDTH_MIN to retrieve the status to determine
if at least one bead section width is out of the min tolerance for a
template.
- New: MbeadGetResult M_STATUS_WIDTH_MAX to retrieve the status to determine
if at least one bead section width is out of the max tolerance for a
template.
- New: MbeadGetResult M_STATUS_INTENSITY_COLOR_MIN to retrieve the status
to determine if at least one bead section color is out of the min
tolerance for a template.
- New: MbeadGetResult M_STATUS_INTENSITY_COLOR_MAX to retrieve the status
to determine if at least one bead section color is out of the max
tolerance for a template.
- New: MbeadGetResult M_INTENSITY_MIN to retrieve the minimum bead
section color value of a template.
- New: MbeadGetResult M_INTENSITY_MAX to retrieve the maximum bead
section color value of a template.
- New: MbeadGetResult M_INTENSITY_MIN_INDEX to retrieve the index of
the bead section with the lowest color value of a template.
- New: MbeadGetResult M_INTENSITY_MAX_INDEX to retrieve the index of
the bead section with the highest color value of a template.
- New: MbeadGetResult M_INTENSITY to retrieve the bead section color
values for a bead template.
- New: MbeadGetResult M_ANGLE to retrieve the bead section angle values.
- New: MbeadGetResult M_TRAINED_INDEX to retrieve the corresponding trained
point index in the trained set of points.
- New: M_FAIL_WARNING_OFFSET control to increase the search box dimension
to find offsetted bead sections beyond the valid offset range.
2.03 Blob module
- New: M_CONVEX_HULL_COG_X and M_CONVEX_HULL_COG_Y blob features.
- New: M_FERET_PERPENDICULAR_TO_MIN_DIAMETER and
M_FERET_PERPENDICULAR_TO_MAX_DIAMETER blob features.
- New: M_FERET_MIN_DIAMETER_ELONGATION and M_FERET_MAX_DIAMETER_ELONGATION
blob features.
- New: M_FERET_AT_PRINCIPAL_AXIS_ANGLE and M_FERET_AT_SECONDARY_AXIS_ANGLE
blob features.
- New: M_ALL_FERETS to retrieve all the calculated ferets.
- New: M_FERET_PRINCIPAL_AXIS_ELONGATION blob feature.
- New: M_RECTANGULARITY blob feature.
- New: M_BLOB_TOUCHING_IMAGE_BORDERS blob feature.
- New: M_BLOB_INCLUSION_STATE result to retrieve whether a specific blob is
included or excluded.
- New: M_TOTAL_NUMBER_OF_RUNS to retrieve the total number of runs in
an image.
- Improvement: improved calculation responsiveness and accuracy when
limiting the processing time using the M_TIMEOUT control.
- Improvement: better use of multiprocessing capabilities.
2.04 Calibration module
- New: McalFixture M_SAME_AS_SOURCE location type to move the relative
coordinate system to the position of the relative coordinate system
of the source.
- New: McalWarp function to define the mapping of a calibration context
based on another calibration context and an optional warping.
- New: McalTransformCoordinate3dList, McalTransformCoordinate, and
McalTransformResultAtPosition M_NO_EXTRAPOLATED_POINTS mode to return
M_INVALID_POINT for pixels transformed outside a piecewise linear
calibrated area.
- New: McalTransformCoordinate3dList now supports 2d calibration contexts
and depth-map buffers through the M_DEPTH_MAP mode.
- New: McalGrid and McalList now work using M_UNIFORM_TRANSFORMATION
calibration contexts.
- Improvement: McalGrid is now more robust using M_CHESSBOARD_GRID grid
type.
- Improvement: M_DISPLACE_RELATIVE_COORD now supports a minimum of 4 points
for the computation to succeed.
- New: McalFixture supports M_RESULT_MET as a location type to fixture from
from the global or a local frame result.
2.05 Code module
- New: M_SIMPLIFIED_CHINESE, M_JAPANESE, M_LATIN, M_KOREAN for
M_STRING_FORMAT.
- Improvement: reading PDF417, TruncatedPDF417 codes at low and very
low speed is more robust in the presence of local deformation or
damaged pattern corners.
- Improvement: reading M_ANY foreground MicroPDF417 is more robust.
- Improvement: reading 4-State codes in the presence of extra bars and/or
uneven spacing is more robust.
- New: M_SEARCH_ANGLE_STEP control used for reading 1D barcodes.
- Improvement: M_TYPE_MIL_ID combination flag is now supported by
McodeInquire.
2.06 Color module
- New: M_HISTOGRAM_VOTE matching technique based on the histogram of
the image.
- New: McolDefine(+M_ADD_COLOR_TO_SAMPLE) to facilitate the definition of a
color sample from multiple color patches.
- New: M_DRAW_SAMPLE_MOSAIC and M_DRAW_SAMPLE_MOSAIC_DONT_CARE drawing
operations for samples defined from multiple color patches.
2.07 Edge Finder module
- Improvement: building the edge chains is more efficient in speed and
memory usage.
2.08 Measurement module
- New: MmeasFindMarker now accepts buffers with regions (MbufSetRegion) to
define the search region.
- New: M_EDGEVALUE_VAR_MIN control to set the minimum expected prominence
of the edge peaks.
- New: M_EDGE_CONTRAST result defined by the difference between the
intensities measured at the edge's maximum of curvature positions.
- New: M_EDGE_CONTRAST_SCORE based on the M_EDGE_CONTRAST result value.
- New: MmeasSetMarker M_CIRCLE_INSIDE_SEARCH_REGION control allows to find a
circle lying partially outside the search region.
- New: M_FIT_ERROR_MAX result is now available for circle markers.
- New: MmeasSetMarker M_CIRCLE_ACCURACY control allows the use of a low
accuracy but typically faster find algorithm.
- New: MmeasSetMarker M_SUBPIXEL_MODE control offers two modes
(M_LOCAL and M_GLOBAL) to determine the subpixel edge location.
- New: MmeasSetMarker M_SEARCH_REGION_CLIPPING_... controls to select the
behavior of the clipping of the search region.
- New: M_EDGE_WIDTH result based on maximal curvature locations, contrary to
M_EDGEVALUE_PEAK_WIDTH, which is based on the threshold.
(see also new M_EDGE_START, M_EDGE_END results).
- New: M_DRAW_EDGEVALUE_PEAK_WIDTH_IN_PROFILE to draw the width of the
measured marker along the edge profile annotation (edge and stripe
markers).
- New: M_DRAW_EDGEVALUE_MIN_IN_PROFILE to draw the threshold levels along
the edge profile annotation (edge and stripe markers).
- New: MmeasDraw with M_RESULT_PER_SUBREGION() macro allows various drawing
operations for annotating the subregions (edge and stripe markers).
- New: MmeasDraw with M_DRAW_POSITION_IN_PROFILE to draw the position of
the edge along the edge profile annotation (edge and stripe markers).
- Improvement: MIL_TYPE_MIL_ID combination flag is now supported by
MmeasInquire.
- New: M_DRAW_INCLUSION_POINT to draw the location of M_INCLUSION_POINT.
- Improvement: improved accuracy of the returned angle when measuring in a
calibrated source image.
2.09 Metrology module
- New: M_DERIVED_GEOMETRY_REGION object and its associated M_REGION_...
controls to define regions from other features (e.g. a ring defined
from three points).
- New: MmetSetRegion(...M_FROM_DERIVED_GEOMETRY_REGION...) to define the
region of a feature from an M_DERIVED_GEOMETRY_REGION object.
- New: MmetSetRegion M_FROM_GRAPHIC_LIST to define a region from a graphic
list object.
- New: MmetDraw M_DRAW_FITTED_EDGELS to draw the edgels participating in the
final fit of a feature.
- New: MmetControl M_EDGEL_TYPE to construct an M_EDGEL feature either the
active edgels of a feature, or from the fitted edgels of a feature.
- New: MmetControl M_GAIN and M_OFFSET to correct for tolerance bias before
validating the result.
- New: MmetInquire M_INDEX_FROM_LABEL to retrieve the index of a feature
or tolerance.
- New: MmetGetResult M_LABEL_VALUE to retrieve the labels of features
and tolerances.
- New: M_COVERAGE values of fitted features can be retrieved.
- New: MmetControl M_REGION_ACCURACY_HIGH can be disabled to allow using
faster but lower accuracy region mapping.
- Improvement: M_EDGEL_SELECTION_RANK now accepts M_DISABLE, which is the
default behavior.
- Improvement: MmetCalculate has additional optimizations when calculating
measured features from rectangular, ring, ring sector, or 1D regions.
- Improvement: has been optimized for multi-core (MP) architectures.
2.10 Model Finder module
- New: MmodControl M_ACTIVE_EDGELS to control the percentage of the most
relevant edgels from the target image that will be used during the find
process, for M_GEOMETRIC contexts.
- New: MmodDraw M_DRAW_ACTIVE_EDGELS to visualize selected edgels from the
target image used during the find process.
- New: M_SEARCH_POSITION_FROM_GRAPHIC_LIST control allows defining the
search region from a graphics list object.
2.11 OCR module
- New: MocrReadString, MocrVerifyString and MocrCalibrationFont functions
now accept buffers with an associated rotated rectangular region
(see MbufSetRegion and MocrControl(M_STRING_ANGLE, M_ACCORDING_TO_REGION))
2.12 Pattern matching module
- New: MpatFindModel and MpatFindModelMultiple now accept buffers with
an associated rotated rectangular region (see MbufSetRegion).
2.13 Registration module
- Improvement: MregTransformCoordinate error message has been clarified
when the result is not calculated.
2.14 String Reader module
- New: The '~' character is now correctly considered as a punctuation
character when added to a font definition.
- New: MstrControl now supports M_RESULT_OUTPUT_UNITS to retrieve result
values in world units. Note that the default setting for the result
output unit is M_ACCORDING_TO_CALIBRATION. Existing calibrated
applications may need to explicitly change this setting.
- Improvement: MstrRead is more robust with strings and fonts containing
punctuation characters.
2.15 3dmap module
- New: "3D Setup Helper.xls" Microsoft Excel sheet, located in the
"Matrox Imaging/Tools/" directory, to help determine the adequate hardware
configuration for a camera-laser setup.
- New: M_Y_THEN_X control to select an alternative gap filling direction.
- New: M_FILL_THRESHOLD_X and M_FILL_THRESHOLD_Y controls to adjust the gap
filling decision of a depth map.
- New: M3dmapDraw M_DRAW_CALIBRATION_PEAKS to draw all the extracted laser
lines during calibration operation.
2.16 Graphic functions
- New: modules can now draw (MmodDraw, MmetDraw...) into a graphics list
object to display the result with zoomable subpixel annotations.
- New: it is now possible to interactively edit a graphics list (see
MdispControl with M_GRAPHIC_LIST_INTERACTIVE and MgraInteractive).
- New: MgraArcAngle M_CONTOUR and M_SECTOR control flag values to specify
an arc style.
- New: MgraControlList M_ARC_STYLE to control the style of an arc in a
graphics list.
- New: MgraControlList M_GRAPHIC_SELECTED control flag to set whether
a given graphic is selected or not.
- New: MgraControl and MgraControlList with M_GRAPHIC_SOURCE_CALIBRATION and
M_FIXTURE to allow drawing annotations under different fixtures.
- New: MgraControl, MgraInquire, MgraControlList, and MgraInquireList, new
control flags for managing interactivity:
M_SPECIFIC_FEATURES_EDITABLE, M_RESIZABLE, M_ROTATABLE, M_MOVABLE,
M_EDITABLE, M_SELECTABLE, M_VISIBLE, M_INTERACTIVE_ANNOTATIONS_COLOR,
M_SELECTED_COLOR, M_SELECTION_RADIUS.
- New: MgraControl new control flags to draw with a zoom and/or an offset:
M_DRAW_OFFSET_X, M_DRAW_OFFSET_Y, M_DRAW_ZOOM_X, M_DRAW_ZOOM_Y.
- New: MgraCopy function to copy one or more graphics from a graphics
list to another.
- New: MgraHookFunction to attach or detach a user function to a graphics
list event such as a modification to the graphics list.
- New: MgraGetHookInfo function to retrieve the information about a
graphics list event.
- New: MgraControlList M_GRAPHIC_CONVERSION_MODE control to determine
how to perform the conversion between world and pixel units of a graphic.
See M_RESHAPE_FROM_POINTS, M_RESHAPE_FOLLOWING_DISTORTION and
M_PRESERVE_SHAPE_AVERAGE.
- New: MgraLines M_INFINITE_LINES control flag to draw infinite lines from
an array of pairs of points.
- New: MgraDraw M_NO_INTERACTIVE_ANNOTATION control flag to disable drawing
interactive annotations.
- New: MgraInquireList M_INTERACTIVE_GRAPHIC_STATE to retrieve the current
state of the interactivity of a graphics list (e.g. IDLE, hover handle)
- New: MgraControlList M_CONSTRAIN_ASPECT_RATIO control flag to set whether
to force the width of objects equal to the height, or not.
- New: MgraControlList M_MULTIPLE_SELECTION control flag to set whether to
enable interactive selection of multiple graphics using the Ctrl key,
or not.
2.17 Regions
- New: MbufSetRegion M_FILL_REGION combination flag to interpret all
graphics list elements as filled graphics.
2.18 Primitives
- New: MbufClearCond function to clear the pixels satisfying a specified
condition.
- New: MimDilate and MimErode M_BINARY_ULTIMATE mode to perform ultimate
morphological dilate and erosion.
- New: MimArith operations: M_EXP, M_CONST_EXP, M_LOG, M_LOG_CONST and
M_CONST_LOG.
- New: MimArith allows many existing operations to be performed interpreting
pixels as logical boolean values (see M_LOGICAL).
- New: MimArith can now perform operations with constant operand in
floating-point accuracy using the M_FLOAT_PROC combination flag.
- New: M_POWER function in MgenLutFunction (A. X ^ B + C).
- New: the following context types can be streamed with MimSave, MimRestore,
and MimStream: M_DEAD_PIXEL_CONTEXT,M_FLAT_FIELD_CONTEXT, M_MATCH_CONTEXT,
M_REARRANGE_CONTEXT and M_STAT_MULTIPLE_CONTEXT.
- New: buffers allocated with the M_LUT attribute now support calls to
MbufControlNeighborhood so that overscan parameters can be set.
- New: MimClip now has the M_SATURATION condition to allow clipping to be
done according to the minimum and maximum of the destination
buffer properties.
- New: MimLocateEvent can now receive M_ALL as a condition to consider
all pixels.
- New: MimLocateEvent has new combination constants: M_LOCAL_MAX_NOT_STRICT,
M_LOCAL_MAX_STRICT_MEDIUM, M_LOCAL_MIN_NOT_STRICT and
M_LOCAL_MIN_STRICT_MEDIUM.
- New: MimConnectMap now supports white overscan. See
MbufControlNeighborhood M_OVERSCAN_REPLACE_VALUE control set to
M_REPLACE_MAX on the received LUT.
- Improvement: the accuracy of the optimized version of MbufBayer with
M_AVERAGE_2X2 interpolation has been improved.
- Improvement: MbufBayer with M_WHITE_BALANCE_CALCULATE now normalizes
coefficients with the maximal average value from the r, g, b bands.
- Improvement: MbufBayer white balancing coefficient calculation is
improved.
- Improvement: the robustness of MimBinarize(...M_TRIANGLE_BISECTION_...) to
small changes in the histogram of the image has been improved.
- Improvement: MimCountDifference is now optimized for SSE2 CPU and for
multiprocessing as well.
2.19 Matrox Profiler
- New: Matrox Profiler tool allows to graphically and interactively analyze
the MIL traces including the MIL functions, the MIL events, the MIL
errors, etc...
3. Deprecated functionalities
3.01 General
- Several MIL constant values (e.g. M_TYPE_CHAR, M_TYPE_SHORT,
M_TYPE_MIL_INT, M_TYPE_MIL_INT32, M_TYPE_MIL_INT64, M_TYPE_MIL_FLOAT,
M_TYPE_DOUBLE, M_TYPE_MIL_ID, M_TYPE_MIL_TEXT, M_AVAILABLE, and
M_SUPPORTED) now require a MIL_INT64 variable to store it, if needed.
- Queuing results by calling several M..GetResult(), M..GetResult1D(),...
with an M_NULL result pointer is now deprecated.
- M_DRAW_RELATIVE_ORIGIN_X, M_DRAW_RELATIVE_ORIGIN_Y, M_DRAW_SCALE_X and
M_DRAW_SCALE_Y controls are now deprecated. They are replaced by
MgraControl on the graphic context with M_DRAW_OFFSET_X, M_DRAW_OFFSET_Y,
M_DRAW_ZOOM_X and M_DRAW_ZOOM_Y.
- MbufControlRegion is deprecated and replaced by MbufControlArea to remove
the ambiguity with the region (MbufSetRegion) functionalities.
3.02 Bead module
- M_POSITION_MODE control is deprecated and replaced by M_TRAINING_PATH.
- M_USER_DEFINED position mode is deprecated and replaced by M_POLYLINE
training path.
- M_AUTO position mode is deprecated and replaced by M_POLYLINE_SEED
training path.
3.03 Calibration module
- McalTransformResult M_ALLOW_INVALID_POINT_OUTPUT combination flag is
deprecated. M_INVALID_POINT is returned when the value cannot be
transformed.
- McalTransformCoordinate, McalTransformCoordinateList,
McalTransformCoordinate3dList, and McalTransformResultAtPosition ;
M_ALLOW_INVALID_POINT_OUTPUT combination flag is deprecated and replaced
by M_NO_POINTS_BEHIND_CAMERA. Note that even if not using
M_NO_POINTS_BEHIND_CAMERA, the function may return M_INVALID_POINT when
transforming pixels on the horizon line.
- McalTransformCoordinate3dList M_PLANE_INTERSECTION mode flag is
deprecated ; use M_DEFAULT instead.
3.04 Color module
- M_CIEDE94_GRAPHIC_ARTS is deprecated and replaced by M_CIE94_GRAPHIC_ARTS.
- M_CIEDE94_TEXTILE is deprecated and replaced by M_CIE94_TEXTILE.
3.05 Measurement module
- M_POSITION_MIN edge result type is deprecated and replaced
by M_EDGEVALUE_PEAK_POS_MIN.
- M_POSITION_MAX edge result type is deprecated and replaced
by M_EDGEVALUE_PEAK_POS_MAX.
- M_CONTRAST edge result type is deprecated and replaced
by M_EDGEVALUE_PEAK_CONTRAST.
- M_CONTRAST_SCORE edge score property is deprecated and replaced
by M_EDGEVALUE_PEAK_CONTRAST_SCORE.
- M_WIDTH edge result type is deprecated and replaced
by M_EDGEVALUE_PEAK_WIDTH.
- M_WIDTH stripe result type is deprecated and replaced by M_STRIPE_WIDTH.
- M_WIDTH_SCORE stripe score characteristic is deprecated and replaced
by M_STRIPE_WIDTH_SCORE.
- The value M_ENABLE for M_SEARCH_REGION_CLIPPING in MmeasSetMarker is
deprecated and replaced by M_MAXIMIZE_AREA.
- M_EDGE_THRESHOLD control is deprecated and replaced by M_EDGEVALUE_MIN.
- The M_ALL parameter index in MmeasDraw is deprecated and replaced
by M_RESULT_ALL_OCCURRENCES.
- The deprecated M_WEIGHT_FACTOR is now obsolete and will lead to a MIL
error. MmeasSetScore must be used instead.
- M_DRAW_ARROW draw type is deprecated and replaced by
M_DRAW_SEARCH_DIRECTION.
3.06 Metrology module
- MmetInquire M_METHOD control flag is deprecated and replaced
by M_OPERATION control flag.
- M_CONSTRUCTION_FEATURE_LABEL is deprecated and replaced
by M_BASE_FEATURE_LABEL.
- M_CONSTRUCTION_FEATURE_INDEX is deprecated and replaced
by M_BASE_SUBFEATURE_INDEX.
- M_NUMBER_OF_CONSTRUCTION_FEATURE_LABEL is deprecated and replaced
by M_BASE_SUBFEATURES_ARRAY_SIZE.
- M_NUMBER_OF_CONSTRUCTION_FEATURE_INDEX is deprecated and replaced
by M_BASE_FEATURES_ARRAY_SIZE.
- M_FILTER_MODE, M_FILTER_WIDTH and M_KERNEL_DEPTH are deprecated.
- M_VISIBLE is deprecated and replaced by M_DRAWABLE.
3.07 Model Finder
- M_FILTER_MODE, M_FILTER_WIDTH and M_KERNEL_DEPTH are deprecated.
3.08 Edge Finder
- M_FILTER_MODE, M_FILTER_WIDTH and M_KERNEL_DEPTH are deprecated.
3.09 Interactive utilities
- All M_INTERACTIVE dialogs are deprecated, except for Save, Restore
and Stream operations.
4. Fixed bugs
4.01 General
- MdigProcess did not propagate its associated calibration to the
grabbed images.
- The M_PROC buffer attribute was not required by all processing
functions including:
MedgeMask, MedgeDraw, MblobDraw, MblobFill, MblobLabel,
MmeasDraw, MmeasFindMarker, MmetDraw, MocrDraw, MregDraw, MregControl,
MstrDraw, MstrExpert, McodeDraw, MpatDraw, MpatAllocModel, MpatCopy,
MpatSetDontCare, McolDefine, McolDistance, McolDraw, McolMask, McolMatch,
McolProject, M3dmapAddScan, M3dmapArith, M3dmapDraw, M3dmapExtract,
M3dmapSetGeometry, M3dmapStat, MbeadDraw, MbeadTrain, MbeadVerify,
MmodControl, MmodDefine, MmodDraw, MmodMask, and MmodFind.
4.02 Bead module
- MbeadTrain, MbeadVerify and MbeadTemplate functions now validates that
only unsigned 8-bit mono buffers are supported.
- MbeadGetResult did not work properly on remote systems when using DMIL.
- MbeadVerify will no longer lead, under some circumstances, to an exception
with an image having no bead.
- MbeadDraw will no longer draw the M_DRAW_SEARCH_BOX centered on the found
position instead of the trained position.
4.03 Blob module
- MblobDraw(...M_DRAW_CONVEX_HULL...) will no longer crash when the
necessary features were not previously selected with MblobSelectFeature().
- MblobGetRuns will no longer crash when using the M_LABELED identification
mode and calculating the convex hull.
- MblobCalculate wrongly accepted binary buffers as the grayscale image,
leading to erroneous results.
- M_RETURN_PARTIAL_RESULTS will no longer return invalid results for some
features such as the general centered moments.
- Inquiring the M_RETURN_PARTIAL_RESULTS will no longer return M_YES/M_NO
but M_ENBALE/M_DISABL as expected.
4.04 Calibration module
- McalTransformCoordinate was not thread-safe.
- McalGrid will no longer fail, in some special cases, when the grid had an
orientation near 45-degrees combined with very high
perspective deformation.
- McalGrid will no longer find wrong positions with M_CHESSBOARD_GRID when
the image is highly distorted.
- McalList did an incorrect parameter check for the minimum number of
3D points needed; it always required 6 points.
- McalFixture was not taking into account the relative coordinate system
when using a calibration context as a reference (instead of an image).
4.05 Code module
- Mix of parity patterns were not properly validated for EAN-13.
- McodeDraw will no longer alter the destination buffer associated
calibration, if any.
- PDF417 and TruncatedPDF417 error correction will no longer fail in
rare situations.
- PDF417 and TruncatedPDF417 accepted impossible combinations of the numbers
of cells in X and Y with X*Y>928, which is greater than the maximum number
of codewords allowed.
- PDF417 and TruncatedPDF417 will no longer detect the wrong number of
columns or rows for some specific codes.
- Size of the 2D part of a composite code will no longer be incorrectly
estimated.
- Under rare circumstances the wrong string will no longer be read
with MicroPDF417: codewords 916 and 917 will no longer be incorrectly
handled.
- Stability of the AIM DPM verification has been fixed for Data Matrix.
- Reading a code in a region outside the target image did not refresh the
M_STATUS of the results.
- McodeRead will no longer fail decoding a degraded QRCode containing the
maximum error correction capacity.
- McodeRead will no longer potentially crash when reading 2-State and
4-State codes.
4.06 Color module
- McolDistance normalization was not working properly for M_BGR32 buffers.
- McolMatch with M_HISTOGRAM_MATCHING directly on a sample image will no
longer lead to an incorrect M_SCORE_RELEVANCE value over 100%.
- McolInquireSafeType (M_MIL_USE_SAFE_TYPE) will no longer lead to false
errors in rare circumstances.
- McolSetMethod wrongly generated an error when using M_HISTOGRAM_MATCHING
with the M_MANHATTAN distance and M_CIELAB color space.
4.07 Edge Finder module
- MedgeGetNeighbors with M_GET_EDGELS will no longer lead to an exception
under some specific circumstances.
4.08 Measurement module
- MmeasSetMarker(...M_RING_RADII..) will no longer lead to an inner radius
greater than the outer radius when the outer radius was set to M_DEFAULT.
- MmeasDraw(...M_DRAW_SEARCH_REGION...) did not draw anything when the
M_BOX_ANGLE control was set to M_ANY. It now draws the box at 0 degree.
- MmeasDraw(...M_DRAW_SEARCH_REGION...) did not draw anything when
M_BOX_SIZE was set to M_DEFAULT.
- MmeasFindMarker stability with circle and stripe markers has been fixed.
- The default search region, defining the whole image, will longer results
in potential unstable calculations.
- The default search region for circle markers was not perfectly at the
image subpixel center.
- MmeasInquire used with the M_SUPPORTED combination flag will no longer
lead to a MIL error.
- Calling MmeasGetResult before MmeasFindMarker will no longer cause
an exception.
- M_CONTRAST result will no longer give negative values when using an M_SHEN
or M_PREWITT filter types.
- MmeasFindMarker will no longer find circles whose centers are not inside
the search region.
- MmeasFindMarker will no longer falsely report a "region out of image"
error message.
- MmeasFindMarker returned inaccurate subedges results if using an
aspect ratio.
- The subpixel location of an edge will no longer be biased in the presence
of very close consecutive edge transitions.
- When using multiple sub boxes foe edge and stripe markers, the data of the
last sub box will no longer be partially projected.
- M_SUB_EDGES_POSITION and M_SUB_EDGES_WEIGHT results will now be ordered
similarly to M_SUB_EDGES_MARKER_INDEX results.
- A specific sequence of MeasSetScore/MmeasFindMarker/MmeasSetScore/
MmeasGetScore calls will no longer result in a crash.
- MmeasDraw with M_DRAW_BOX no longer leads to an incorrect drawing when
some box settings are set to M_DEFAULT.
- All edge and stripe marker annotations (crosses) are all consistently
drawn at the angle of the measurement box.
- MmeasStream will no longer generate a small memory leak.
- The M_INCLUSION_POINT will no longer ignore the input unit setting.
4.09 Metrology module
- When drawing in a calibrated destination, the Y-axis of the local frames
was not aligned with the world Y-axis, but orthogonal to the world
aligned X axis.
- MmetCalculate will no longer skip calculations if only the calibration
information of an image changed and not its content.
- MmetDraw with M_ACTIVE_EDGELS will no longer draw the fitted edgels
instead of the selected ones.
- MmetDraw, horizontal and vertical constructed parametric lines could not
be drawn.
- The arrow of an arc region was not drawn properly when the start angle was
greater than the end angle.
- MmetDraw of edgels did not properly consider the calibration of the image
used by MmetCalculate.
- MmetDraw will no longer draw unsuccessful calculated features at (0, 0).
It now draws nothing.
- MmetDraw will no longer ignore the calibration information of the
destination image when drawing tolerances.
- The farthest point (M_MAX_DISTANCE_POINT) between a circle and an arc will
no longer give the wrong point.
- MmetGetResult will no longer return the wrong pixel unit coordinates for
measured points when calculating on a calibrated image.
- M_RESULT_OUTPUT_FRAME set to M_GLOBAL_FRAME did not return the
correct results.
- Perpendicularity and parallelism tolerances were not always returning the
smallest angular values.
- The construction of a perpendicular line, a parallel line, or a line at
angle was not properly validating the type of the reference feature. An
error message will now pop when the construction is invalid.
- MmetCalculate will no longer find edgels outside the measured region when
using a calibration with a very distorted pixel to world mapping.
4.10 Model Finder module
- MmodFind will no longer generate a run-time exception in very rare
configuration with very short extracted edge chains.
- Under some circumstances, MmodFind could not find a synthetic rectangular
model if M_ACCURACY was set to M_LOW.
- MmodRestore(... M_WITH_CALIBRATION ...) will no longer lead to an error
in MmodPreprocess.
- Robustness fixes with M_GEOMETRIC_CONTROLLED contexts when using
the M_LAST_LEVEL control.
- MmodDraw M_DRAW_MODEL_IMAGE and M_DRAW_EDGES for synthetic and CAD models
did not take into account the graphic context color setting, but forced
the color to 255.
4.11 OCR module
- M_CHAR_SPACING result returned incorrect values when the read string was
at an angle.
- MocrDraw ignored the color of the received graphic context.
- MocrGetResult with M_CHAR_SPACING, M_CHAR_SIZE_X/Y, M_STRING_ANGLE was
not returning results in world units even if the read was performed with
a calibrated image.
- MocrPreprocess will no longer lead to an exception if
M_TARGET_CHAR_SPACING is set lower than 2.
- MocrReadString will no longer give unstable results when there are fewer
characters in the image than the number of characters to read.
- MocrReadString was not working properly when reading characters at angle
with a different scale than the font.
- MocrReadString will no longer lead to an exception when
M_STRING_ANGLE_DELTA_NEG or M_STRING_ANGLE_DELTA_POS are set to 2.
- Predefined SEMI font contexts will no longer read with a bad checksum if
the number of strings is greater than one.
4.12 Pattern matching module
- MpatFindModel was not working perfectly with very large buffers
(one pixel size over 65535).
- MpatFindModel with circular overscan model will no longer find multiple
occurrences at the same position with multiprocessing enabled.
- MpatFindModel will no longer have a memory leak with multiprocessing
enabled.
- MpatFindModel is now able to find a model if the occurrence position
is too close to the search region borders.
- M_MIN_SPACING_X, M_MIN_SPACING_Y were not working properly when searching
for a model at an angle.
- M_SAVE_SUMS set to M_ENABLE will no longer lead to incorrect results when
the center of the model is also changed.
- MpatFindModel will no longer crash when setting an M_MODEL_STEP to
an unexpected value.
- MpatFindModel is now able to find large models with don't care mask and
model step of 2.
- In rare cases, MpatFindModel will no longer crash when using very large
models.
4.13 Registration module
- MregTransformImage erroneously accepted destination buffer formats other
than 8+M_UNSIGNED.
4.14 String Reader module
- For MstrRead and MstrExpert, a lack of memory will no longer lead to a
memory leak in very special circumstances.
- MstrRead will no longer be unable to read in an image with an associated
region and with the M_THICKEN_CHAR control set to a non-null value.
- MstrRead will no longer potentially crash or lead to incorrect results
when disabling some characters from a fontless context.
4.15 3dmap module
- M3dmapExtract M_FILL_MISSING_DATA_ONLY was not allowed on empty results.
- M3dmapExtract with M_USE_DESTINATION_SCALES now works on physically
corrected images with a pixel rotation (fixtured buffers).
- M3dmapTriangulate now works with DMIL remote systems.
- M3dmapArith did not work properly on two source images with different
calibration child offsets. It now reports an error.
- M3dmapExtract with M_USE_DESTINATION_SCALES did not take into account
the calibration child offsets when extracting from a child of the
3d-corrected buffer.
- M3dmapExtract with M_USE_DESTINATION_SCALES did not work properly with
a physically corrected image destination having a rotated relative
coordinate system (fixture).
- M3dmapCalibrate in M_CALIBRATED_CAMERA_LINEAR_MOTION mode will no longer
ignore the M_Y_AXIS_UP setting of the camera calibration.
- M3dmapExtract will no longer ignore to set the relative coordinate system.
The relative coordinate system is now correctly set to the 3d
reconstruction coordinate system.
- M3dmapArith will no longer ignore to set the relative coordinate system.
The relative coordinate system is now correctly set to the 3d
reconstruction coordinate system.
4.16 Graphic functions
- MgraLines did not always render the exact same line when executed on
child buffers versus when executed on its ancestor buffer.
- MgraArcFill will no longer crash for very large arcs.
- MgraInquireList(...M_LAST_LABEL...) will no longer erroneously return a
zero label value.
- MgraClear will no longer ignore the colors from the received graphic
context with on-board allocated buffers.
- Mgra functions execution will no longer be abnormally slow on GPU systems.
- MgraControlList on M_ALL or M_ALL_SELECTED objects will no longer report
an error when the control does not apply to some graphic objects.
- MgraControlList did not correctly handle the M_AVAILABLE, M_SUPPORTED, and
M_DEFAULT combination flags.
- MgraText will now correctly interpret the M_FONT_SIZE setting.
- After deleting the last graphic of a graphic list, inquire to M_LAST_LABEL
will no longer return an invalid label value but M_NO_LABEL instead.
4.17 Regions
- MbufSetRegion function will no longer lead to an exception if function
calls to the calibration module are not previously done.
4.18 Primitives
- Loading or streaming an image with no associated region into a buffer
was deleting its region, if any. The region now remains intact.
MbufSetRegion(...M_DELETE...) must be used to remove a region.
- MbufBayer applied the gamma correction after the demosaicing operation.
- MbufBayer with M_AVERAGE_2X2 will no longer lead to an exception with
very small buffers.
- MimMorphic with M_BINARY processing mode used uninitialized overscan data
when the destination was an 8-bits buffer.
- MimMorphic with 0 iterations had inconsistent behaviors. It now copies the
source to the destination.
- MimMorphic performed all the iterations except the first one in the data
type of destination. All iterations are now done and saturated according
to the data type of the source buffer, then the result is put to the
destination buffer.
- MimMorphic was truncating intermediate results when either the source,
the destination or the structuring element was a floating-point buffer.
- MimThin and MimThick with 0 iterations had inconsistent behaviors
regarding MimMorphic. It now copies the source to the destination.
- MbufCopyColor2d was allowed for subsampled bands (e.g. U and V bands for
subsampled YUV formats). It is now protected.
- MimTransform with M_FFT will no longer give inexact results with some
images when doing a reverse transform using only the 'real' part of
the results.
- MimDraw is now working properly on 3-band color buffers.
- MimDraw now correctly considers the color set by the M_DEFAULT
graphic context.
- MimDraw with M_DRAW_PEAKS will no longer give annotations off by half
a pixel.
- MimDilate, MimErode, MimThin and MimThick functions always perform a
buffer copy if the number of iterations is set to 0.
- MimPolarTransform will no longer lead to an exception when using a
small angle.
- MimConvert will no longer give erroneous results with M_BGR32+M_PACKED
image buffers when using matrix multiplication with a 4x3 matrix.
- MimLocateEvent will no longer lead to an exception when using
multiprocessing optimization.
- MimMatch will no longer lead to an exception when using very large models.
- MimPolarTransform will no longer have inconsistent results when doing
a transform and its reverse.
- Copying a YUV source image to an RGB destination with a type equal to
or greater than 16 bits no longer leads to erroneous values.
- MimPolarTransform will no longer leads to erroneous values when the
buffer is a child band of a non-regular buffer.
- MimThin with M_BINARY2 and a white overscan no longer occasionally leads
to a crash.
- Under Windows XP, MappInquireMP(M_MEMORY_BANK_AFFINITY_MASK) and
MbufAllocColor will no longer be inconsistent.
- MbufExport(...M_WITH_CALIBRATION...) will no longer malfunction with
compressed buffers.
4.19 Interactive utilities
- Code Reader interactive utility will no longer prevent you from properly
editing the number of cells in X with PDF417 and MicroPDF418 codes.
- Code Reader interactive utility will no longer lead to an exception when
reading many code symbols.
- OCR Reader interactive utility did not report read, verify of
preprocessing errors.
- OCR Reader interactive utility did give the focus to an inaccurate
window when selecting a result.
- Color Analysis interactive utility will no longer display some artifacts
when zooming the display of a color sample.
- Model Finder interactive utility did not correctly update the preprocess
status when modifying the pixel scale of a synthetic model.
- Metrology interactive utility will no longer lead to an exception when
opening the 'modify' dialog on a tolerance defined with a single feature.
- Metrology was not correctly displaying a reference template images when
using 32+M_FLOAT format buffer.
- The calibration dialog in the interactive utilities erroneously allowed
associating an image document to an empty calibration, leading
to MIL errors.
- Using an interactive utility with a remote Iris system (DMIL) and
with no local license prevented the deinterlace and the calibration
dialogs from working properly.
- Pasting an image in an interactive utility gave an inaccurate error
message when no license was present.
- Stability fixes in the continuous grab functionality.
- Stability fixes in the calibration dialog when using the
'Associate to digitizer' option.
- Opening the MIL help will no longer lead to an incorrect section if
MIL help was already opened.
5. Known limitations
5.01 Graphic functions
- Interactive edition only tracks the region distortion, but not the
drawing. The region drawing can track the region distortion when the
M_CONVERSION_MODE is set to M_RESHAPE_FOLLOWING_DISTORSION.
5.02 Interactive utilities
- The MIL Interactive utilities (e.g. Matrox Inspector) ignore regions
of images.
- M_DRAW_OFFSET_X/Y and M_DRAW_ZOOM_X/Y controls are currently not used
by the interactive edition nor by the processing modules when using
M_VECTOR type regions.
- Calibration dialog of the specific MIL Interactive utilities (e.g.
Model Finder GUI): if a child buffer is selected, the calibration
is performed using the child, but associated to the parent. This
leads to incoherent results. Selecting a child should be avoided
when performing a calibration using a specific MIL Interactive
utilities.
- Inspector Measurement dialog has been updated to prevent from using
the old deprecated scores.
5.03 Regions
- The raster version of a MIL 10 image's region is discarded when the
image is loaded in an older version of MIL (i.e. MIL 9 PP2 and older).
Forcing MbufSetRegion(...M_RASTERIZE...) after loading the image will
restore the raster version.
5.04 Metrology
- Contexts having features or tolerances with an associated name
(MmetSetname) and saved under Windows, cannot be restored on Linux; and
vice-versa. Removing the names solves the issue.
5.05 Matrox Profiler
- Combining application and thread level activation or deactivation of the
trace log can lead to incoherent results when visualized with the Matrox
Profiler.
5.06 Streaming functions
- Loading a stream (file or memory) containing a different object type when
using MxxxStream with an M_LOAD operation will not generate a MIL error
as it should. The loaded object could be in an unstable state,
particularly on a remote DMIL system.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 7: Differences between MIL 9 Processing Pack 2 with Update 56 and
MIL 9 Processing Pack 2
Table of Contents for Section 7
1. Overview
2. Fixed bugs
2.1 Pattern Matching
-------------------------------------------------------------------------------
1. Overview
- MIL 9 Update 56 includes important bug fixes to the MIL Pattern Matching
module.
- MIL 9 Update 56 is an update to MIL 9 with Processing Pack 2 only.
2. Fixed bugs
2.1 Pattern Matching
- MpatFindModel: a memory leak, when using circular overscan with MIL
MP (multi-core processing) enabled, has been fixed.
- MpatFindModel: could return multiple occurrences at the same location
when using circular overscan with MIL MP (multi-core processing) enabled.
- After setting the advanced M_MAX_INITIAL_PEAKS parameters, the number of
occurrences returned by the module was always 1.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 8: Differences between MIL 9 Processing Pack 2 with Update 45 and
MIL 9 Processing Pack 2
Table of Contents for Section 8
1. Overview
2. New functionalities and improvements
2.1 Code reading
2.2 Code verification
3. Fixed bugs
3.1 Code reading
3.2 Code writing
3.3 Code verification
4. Known issues
-------------------------------------------------------------------------------
1. Overview
- MIL 9 Update 45 includes new functionalities, performance optimizations
and general improvements for the MIL Code Reader module.
The key features are:
* MicroQR code support;
* More distortion support for Data Matrix, QR and 4-State codes.
* Verification of Data Matrix, QR and MicroQR codes using ISO 15415 and
AIM DPM-1-2006 standards;
- MIL 9 Update 45 is an update to MIL 9 Processing Pack 2 only.
- MIL 9 Update 45 updates the MIL Help files accordingly.
2. New functionalities and improvements
2.1 Code reading
- New support for reading Micro-QR codes
- QR codes: M_DISTORTION control type supports M_UNEVEN_GRID_STEP to read
symbols with a moderate amount of distortion between alignment patterns.
- QR codes: M_FOREGROUND_VALUE control type supports M_FOREGROUND_ANY.
- QR codes: supports character sets as defined in ISO/IEC 18004:2006.
- Data Matrix: M_DISTORTION control type supports the new flag
M_PERSPECTIVE_UNEVEN_GRID_STEP to read symbols distorted by a combination
of uneven grid steps and an affine transformation e.g. (perspective,
shearing, etc.)
- MicroQR codes: supports character sets as defined in ISO/IEC 18004:2006.
- Data Matrix: improved localization performance especially for rotated
codes.
- 4-State codes: improved robustness for severely wavy distorted codes.
- PDF417: improved robustness when facing tight ROIs or small quiet zones.
- Truncated PDF417: improved robustness when facing tight ROIs or small
quiet zones.
- Pharmacodes: increased robustness for out-of-specification bar
size variations.
- New M_SEARCH_ANGLE_STEP control type for 1D codes.
- New M_CHECK_QUIET_ZONE control type to enable/disable assuming the
existence of a valid quiet zone around the codes.
- New M_STRING_FORMAT control type, supported on Windows platforms except
CE.
- New M_STRING_FORMAT control type: M_SIMPLIFIED_CHINESE for outputting
decoded strings in Simplified Chinese (Windows 936) code page, for
M_PDF417, M_MICROPDF417, M_TRUNCATED_PDF417, M_QRCODE, M_MICROQRCODE,
and M_DATAMATRIX.
- New M_STRING_FORMAT control type: M_JAPANESE for outputting decoded
strings in Japanese (Windows 932) code page, for M_PDF417, M_MICROPDF417,
M_TRUNCATED_PDF417, M_QRCODE, M_MICROQRCODE, and M_DATAMATRIX.
- New M_STRING_FORMAT control type: M_LATIN for outputting strings in Latin
(Windows 1252) code page, for M_PDF417, M_MICROPDF417, M_TRUNCATED_PDF417,
M_QRCODE, M_MICROQRCODE, and M_DATAMATRIX.
- New M_STRING_FORMAT control type: M_KOREAN for outputting strings in
Korean (Windows 949) code page, for M_PDF417, M_MICROPDF417,
M_TRUNCATED_PDF417, M_QRCODE, M_MICROQRCODE, and M_DATAMATRIX.
2.2 Code verification
- New AIM DPM-1-2006 standard support for Data Matrix, QR and MicroQR codes.
- Complete ISO 15415 standard support for Data Matrix, QR and MicroQR codes.
- New M_VERIFICATION_STANDARD control type to select the desired
specification to use for Data Matrix codes, QR codes, and Micro-QR codes.
The possible values are: M_ISO_VERIFICATION, and M_AIMDPM_VERIFICATION.
- Deprecation and potential change of behavior: the old ANSI verification
is deprecated and replaced, by default, by the ISO verification standard.
- New M_NUMBER_OF_SCANS_PER_ROW control type to set the number of scans per
row for the verification of linear codes and cross-row codes.
- New M_INSPECTION_BAND_RATIO control type to set the inspection band ratio
for the verification of linear codes and cross-row codes.
3. Fixed bugs
3.1 Code reading
- PDF417 code: start/stop pattern detection could not work properly.
- PDF417 code: could fail to correctly determine the number of columns
and rows.
- Planet code: could fail to correctly determine the module size.
- Postnet code: could fail to correctly determine the module size.
- 4-State code: could fail to correctly determine the module size.
- PDF417 code: the error correction could fail when the number of
codewords was large.
- 2D code drawings: a regular grid was drawn instead of the fetched
positions.
3.2 Code writing
- Australian 4-State code: fixed crash when encoding raw data.
- EAN14 code: fixed potential heap corruption.
- PDF417 code: code not generated correctly: the symbol length
descriptor was not computed correctly.
3.3 Barcode verification
- MicroPDF417 code: fixed potential Row Address Pattern (RAP) grade; it
cannot be larger than 52.
4. Known issues
4.1 It is not recommended reinstalling MIL 9 Processing Pack 2 after
MIL 9 Update 45: this will replace some files required for the proper
functioning of this update.
4.2 When upgrading processing specific example projects from Visual Studio
2005 to Visual Studio 2008 or Visual Studio 2010 with UAC enabled,
there will be a warning in the conversion log indicating that current
user privileges have been added to the project. This will not cause
any problems.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 9: Differences between MIL 9 Processing Pack 2
and MIL 9 Processing Pack 1
Table of Contents for Section 9
1. Overview
2. New processing functionality and improvements
2.1. Processing specific examples (New!)
2.2. Bead module (New!)
2.3. Calibration module (New fixturing!)
2.4. Color module
2.5. 3dmap module
2.6. Metrology module
2.7. Code Reader module
2.8. Blob module
2.9. Measurement module
2.10. Pattern Matching module
2.11. Model Finder module
2.12. Edge Finder module
2.13. String Reader module
2.14. OCR module
2.15. Registration module
2.16. Graphic functions
2.17. Mim and general processing functions
2.18. Interactive utilities
3. Fixed bugs
3.1. Code reader module
3.2. Blob module
3.3. Calibration module
3.4. String Reader module
3.5. Edge Finder module
3.6. Model Finder module
3.7. Color module
3.8. Metrology module
3.9 3dmap module
3.10 Pattern matching module
3.11 Measurement module
3.12 Registration module
3.13 OCR module
3.14 Graphic functions
3.15. Mim and general functions
3.16. Interactive utilities
4. Mim functions and performance optimizations
5. Board specifics
6. Miscellaneous
-------------------------------------------------------------------------------
1. Overview
- MIL 9 Processing Pack 2 includes new processing functionalities,
performance optimizations and general improvements, such as: a new bead
inspection module, a GUI for the color module, new fixturing capability,
world input unit support, enhanced 3dmap capabilities,... and much more!
- MIL 9 Processing Pack 2 also installs ActiveMIL controls. For more
information, see the ActiveMIL readme file.
- MIL 9 Processing Pack 2 incorporates and updates the display
functionality beyond that introduced by MIL 9 Update 16.
- MIL 9 is the minimum requirement for this Processing Pack 2.
2. New processing functionality and improvements
1. Processing specific examples (New!)
The new %MIL_PATH%\Examples\Processing-Specific directory includes
MIL processing specific examples (frameworks, utilities, applications...)!
- Preprocessing: SuperResolution, Morphology
- Pattern Matching: PCBModelMatching
- Measurement: CircleMeasurement, AdvancedMeasurement
- Color Inspection: FoodInspection_Cereal
- Code: VariousCodeReading, FixturedCodeRead, DotSpacing, CodeGrading
- CameraLaser3D: PhotonFocus, SickRangerC, MatroxGatorEye
- 3dRobotic: CameraOnRobotArmCalibration
- 3dReconstruction: CameraLaserCalibration, StereoCalibration
2- Bead module (New!)
- New: module for inspecting beads such as sealants or adhesives. It
calculates the bead's width and the bead's positions, and identifies
out-of-range bead's properties and gaps.
3- Calibration module
- New: Fixturing capability in MIL.
- New: McalFixture API to move or to learn a relative position using either
user-defined values or result values, such as Model Finder results
or Pattern Matching results; this allows you to further fixture subsequent
operations.
- New: M_..._INPUT_UNITS controls to set in world/pixel input units most of
the module setting values (blob, code, measurement, bead,...)
- New: McalTransformImage supports M_OVERSCAN_FAST interpolation mode.
- New: McalRelativeOrigin and McalSetCoordinateSystem can now be
set to a calibrated image; this allows you to further fixture subsequent
operations.
- New: McalTransformResultAtPosition allows you to accurately transform
from/to world/pixel angles values at a given location.
- New: M_CONSTANT_PIXEL_SIZE flag to inquire whether a calibrated image has
a constant pixel size.
- Improvement: M_Y_AXIS_UP can now be inquired on a calibrated image.
- New: McalUniform API to associate a default uniform calibration to an
image image and to set its scale, rotation, and translation properties;
equivalent inquiries added to retrieve information about translation,
rotation, and scale values.
- New: M_WITH_CALIBRATION flag allows an associated calibration to be
saved with its owner such as a buffer, a model finder context or a
metrology context.
- New: McalDraw supports new operations to draw the coordinate systems, with
optionnal axis, frames, tick marks, and legends.
- New: M_TRANSFORM_IMAGE_FILL_MODE is now deprecated and replaced by the use
of combination flags in McalTransformImage.
- New: M_USE_DESTINATION_CALIBRATION McalTransformImage combination flag;
this allows you to further fixture subsequent operations using corrected
images.
- New: M_OUTPUT_UNITS has been deprecated in McalControl and replaced by
the M_RESULT_OUTPUT_UNITS controls in modules.
- Improvement: better M_MATHEMATICAL_EXCEPTION protection when unstable
configuration of calibration points are submitted.
- Improvement: it is now possible to pass an index to McalDraw when using
M_DRAW_WORLD_POINTS or M_DRAW_IMAGE_POINTS on a calibrated image.
- New: McalRelativeOrigin(M_COMPOSE_WITH_CURRENT) receives coordinates in
the relative coordinate system.
- New: McalDraw(M_DRAW_FIXTURING_OFFSET) to draw the fixture offset.
- Improvement: McalInquire(... M_Y_AXIS_UP...) is now available on image
buffer.
4- Color module
- New: McolDistance and McolProject have been optimized using SIMD
instructions.
- New: McolDistance and McolProject have been optimized for multi-core (MP)
architectures.
- New: McolMatch with the vote method has been optimized using SIMD
instructions and for multi-core (MP) architectures.
- New: McolMatch histogram based method useful for matching
multi-colored areas.
- New: advanced inquires about color sample information and statistics.
- New: support for additional CIE distance types (M_CMC_PERCEPTIBILITY,
M_CMC_ACCEPTABILITY, M_CIEDE94_GRAPHIC_ARTS, M_CIEDE94_TEXTILE,
and M_CIEDE2000).
- New: McolDraw, M_INVERTED_COLORS combination flag for drawing LUT and
result images using inverted colors.
5- 3dmap module
- New: uniform calibration automatically associated for generating
calibrated depth-map.
- New: automatic scaling for corrected depth map.
- New: it is now possible to call multiple M3dmapExtract() on a same
destination buffer, and to use this destination buffer as
"3D point accumulator" by setting the M_EXTRACTION_CUMULATIVE control of
the result buffer.
- New: M_CLEAR_DATA control to clear the result buffer data while leaving
the current position unchanged.
- New: M3dmapSetGeometry function to fit a surface on a depth map or
to define a geometry using analytic parameters in world units.
- New: M3dmapStat function to compute a variety of statistics on the depth
map image, such as volume or deviations. Some statistics can also be
measured relative to a reference geometry (see M3dmapSetGeometry).
- New: M3dmapArith function to perform point-to-point arithmetic between
depth maps and/or reference geometries (see M3dmapSetGeometry).
- New: M3dmapDraw to draw various 3dmap context or result information or
reference geometries.
- New: M3dmapSave, M3dmapRestore, M3dmapStream to save, restore, and stream
3dmap contexts.
- New: M_LASER_CONTEXT_TYPE flag to inquire about the context type
(M_DEPTH_CORRECTION or M_CALIBRATED_CAMERA_LINEAR_MOTION).
- New: inquires and drawings to help the diagnostic of the result of the
calibration process, such as the number of columns with missing data,
the plane parameters, the calibration plane fit error,...
- New: M_REMOVE_LAST_SCAN control to undo last call to M3dmapAddScan.
6- Metrology module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: M_EDGEL_SELECTION_RANK control to specify which edgel position to
select relative to the orientation of rectangular search regions.
- New: segment M_CONSTRUCTION from arc/circle centers and/or local frames.
- New: local frame M_CONSTRUCTION from arc/circle centers and/or local frame
- New: circle M_CONSTRUCTION from arc/circle centers and/or local frames.
- New: arc M_CONSTRUCTION from arc/circle centers and/or local frames.
- Improvement: of the inner circle fit calculation.
- New: MmetName API to set/get names to features and tolerances.
- New: M_DRAW_NAME combination flag to draw feature and tolerance names.
- Improvement: the M_DRAW_REFERENCE_FEATURES flag is deprecated and is
replaced by the M_DRAW_TOLERANCE_FEATURES flag.
- New: M_NUMBER_OF_CONSTRUCTION_FEATURE_LABEL to retrieve the number of
features used to define a constructed feature or a tolerance.
- New: M_NUMBER_OF_CONSTRUCTION_FEATURE_INDEX to retrieve the number of
features indices used to define a constructed feature or a tolerance.
- New: M_CONSTRUCTION_FEATURE_LABEL(i) to retrieve the label of the i-th
feature to define a constructed feature or a tolerance.
- New: M_CONSTRUCTION_FEATURE_INDEX(i) to retrieve the label of the i-th
feature index to define a constructed feature or a tolerance.
- New: M_METHOD to retrieve the method used for a constructed feature.
- New: M_DEPENDENCIES to inquire if a feature has some other features
or tolerances derived or based on it.
- New: M_MODIFY MmetAdd... flag to modify the construction properties of an
already existed feature or tolerance.
- New: MmetSave/Restore/Stream support M_WITH_CALIBRATION.
7- Code Reader module
- New: M_DISTORTION control to enable reading distorted codes.
Currently supported is the M_UNEVEN_GRID_STEP distortion mode for
Data Matrix codes.
- New: supports for 4-States (also known as IMB) barcodes with US, UK,
Australian C and N encodings.
- New: the flag name M_RSSCODE is deprecated and replaced by the flag
name M_GS1_DATABAR.
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: M_..._INPUT_UNITS control to set input settings such as the cell
size, the dot spacing, the angle, the aperture size, ... in world or
pixel units.
- New: McodeRead supports input regions (see MbufSetRegion).
- New: supports for EAN14 and GS1-128 encodings.
- New: M_STRING_FORMAT control to set the output string format when
applicable. Currently supported is the M_UPCE_COMPRESSED mode for
UPC-E codes; M_GS1_HUMAN_READABLE for output in human readable (applied
only on M_CODE128, M_GS1_128, M_EAN14, M_GS1_DATABAR, M_COMPOSITECODE
and M_DATAMATRIX); and M_AUTO_FORMAT for default recommended output format.
- New: M_DATA_CODEWORDS result flag to retrieve the data code for most of
the composite and 2D code types.
- Improvement: of the localization for 1D barcodes.
- Improvement: of the position accuracy when reading multiple codes.
- Improvement: MicroPDF417 start/stop grade symbol detection.
- Improvement: the reading of multiple codes with speed = M_VERY_LOW.
- New: McodeRead and McodeVerify take advantage of the new MbufSetRegion
function used to set a region that defines the pixels of interest
for an image.
- New: support for GS1 Data Matrix.
8- Blob module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: M_BOX_AREA feature to determine the blob's bounding box area.
- New: M_BOX_ASPECT_RATIO feature to determine the ratio between X and Y
dimensions of the blob's bounding box.
- New: M_BOX_FILL_RATIO feature to determine the ratio between the blob's
area and its bounding box area.
- New: M_CONVEX_HULL_FILL_RATIO feature to determine the ratio between the
blob's area and its convex hull area.
- New: M_CONVEX_HULL_X and M_CONVEX_HULL_Y to determine the X and Y
coordinates of the convex hull perimeter.
- New: M_NUMBER_OF_CONVEX_HULL_POINTS feature to determine the number of
point of the convex hull.
- New: M_CONVEX_HULL_AREA feature to determine the blob's convex hull area.
- New: MblobCalculate additional optimizations using SIMD instructions for
calculating the blob's RLE.
- New: M_RETURN_PARTIAL_RESULTS control flag to enable retrieving the
partially calculated blobs when a stop criterion condition is met
(i.e. when the M_MAX_BLOBS or M_TIMEOUT condition is met).
- New: MblobCalculate takes advantage of the new MbufSetRegion function used
to set a region that defines the pixels of interest for an image.
- New: M_WORLD_BOX, M_WORLD_FERET_X and M_WORLD_FERET_Y features and
corresponding drawings to determine the features in the world system when
using calibration.
9- Measurement module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: circle markers (M_CIRCLE) are now supported.
- New: MmeasSetScore API to define advanced edge selection rules.
- New: M_..._INPUT_UNITS flags to set input settings in world or pixel units
- New: M_SEARCH_REGION_CLIPPING control. When enabled, an optimal box will
be computed when the search box falls partially outside the image limits.
- New: M_SEARCH_REGION_WAS_CLIPPED result to retrieve whether a measurement
was performed using a clipped measurement box.
- New: M_FOUND_BOX_ANGLE status result can now be retrieved.
- New: M_POSITION_IN_MARKER result value to retrieve the distance relative
to the origin of the measurement box of an edge or a stripe occurrence.
- New: M_FIT_ERROR_MAX result to retrieve the maximum distance bewteen
the fitted geometry and the transition positions found in measurement
sub-regions for an edge, stripe, or circle marker.
- New: M_MAX_ASSOCIATION_DISTANCE control to reject outliers when fitting
the marker on sub-box positions.
- New: M_NUMBER_OF_OUTLIERS result to retrieve the number of transition
positions found in measurement sub-regions discarded by the fit process
for an edge, stripe, or circle marker.
- New: M_LINE_END_POINT_FIRST and M_LINE_END_POINT_SECOND results.
- Improvement: some additional optimization have been performed
when using sub-boxes.
- New: world unit line equation (A,B,C) parameters.
- Deprecation: M_POSITION_INSIDE_STRIPE flag has been renamed to
M_INCLUSION_POINT_INSIDE_STRIPE and is used in conjunction with
the M_INCLUSION_POINT flag.
10- Pattern Matching module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: the limitation of the max number of models allowed (i.e. 15)
for MpatAllocAutoModel has been removed.
- New: M_SEARCH_ANGLE_TOLERANCE control can now be set to M_AUTO. If set to
M_AUTO, the tolerance is optimally determined regarding the angular model
auto-correlation and the new M_ROTATED_MODEL_MINIMUM_SCORE control value.
- New: M_FIRST_LEVEL control can now be set to either M_AUTO_SIZE_BASED or
to the new M_AUTO_CONTENT_BASED mode, useful for but not limited to large
models composed of small features.
- New: M_MAX_INITIAL_PEAKS control flag which can be set to M_ALL to force
an exhaustive analysis of the correlation peaks when a very large set of
potential occurrences (>100) is present in the target image.
- New: maximum first and last level limitation (i.e. 7) has been removed,
and the new M_MODEL_MAX_LEVEL flag allows inquiring about the max possible
level for a given model.
- New: MpatFindModel now supports 1-D target images.
- New: MpatFindModel has been optimized for multi-core (MP) architectures.
- New: MpatFindModel has been optimized using SIMD instructions for large
models.
11- Model Finder module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: low accuracy control enables using a less accurate but faster method
to compute the edgels' angle.
- New: MmodFind can now be done with a mix of uncalibrated models/target
and uniform calibrated scale=1 models/target.
- New: saving/restoring model finder context with M_WITH_CALIBRATION.
12- Edge Finder module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: M_ANGLE_ACCURACY accuracy control. Setting M_LOW enables a less
accurate but faster method to compute the edgels' angle.
13- String Reader module
- New: MstrRead supports input regions (see MbufSetRegion)
- New: MstrRead takes advantage of the new MbufSetRegion function used
to set a region that defines the pixels of interest for an image.
14- OCR module
- New: M_RESULT_OUTPUT_UNITS to retrieve results in world or pixel units
(replaces M_OUTPUT_UNITS in McalControl).
- New: M_EXTRA_CHARACTERS control, which is useful when the target image
string contains more characters than the expected characters to be read.
15- Registration module
- Improvement: the registration module is now robust to not-enough-memory
exceptions.
16- Graphic functions
- New: MgraAllocList, MgraControlList, MgraInquireList, and MgraDraw, which
enables vectorial capabilities in MIL.
- New: MgraLines capability to draw Polygon and Polyline geometries.
- New: M_TEXT_ALIGN_HORIZONTAL/VERTICAL control to set the text alignment.
- New: M_INPUT_UNITS control flag to support annotation capabilities in
world / pixel units.
- New: M_ASSOCIATED_GRAPHIC_LIST_ID MdispControl flag to associate a graphic
list to a MIL display object.
- New: M_UPDATE_GRAPHIC_LIST (M_ENABLE/M_DISABLE, MdispControl) to set
whether or not modifications applied to the graphic-list associated to
the display must be updated to the display.
- New: M_REGION_INSIDE_COLOR, M_REGION_INSIDE_SHOW, M_REGION_OUTSIDE_COLOR,
and M_REGION_OUTSIDE_SHOW display controls to set the buffer region
display properties.
- New: M_MOUSE_LEFT_BUTTON_DOWN, M_MOUSE_RIGHT_BUTTON_DOWN,
M_MOUSE_LEFT_BUTTON_UP, M_MOUSE_RIGHT_BUTTON_UP, M_MOUSE_MOVE display hook
types to hook on mouse events.
- New: M_MOUSE_POSITION_X, M_MOUSE_POSITION_Y, M_MOUSE_POSITION_BUFFER_X,
and M_MOUSE_POSITION_BUFFER_Y flag to retrieve the information about
the mouse position.
- New: display overlay now automatically follows the calibration of
the selected buffer, when calibrated.
- New: Exclusive(M_EXCLUSIVE) displays can be set to the current desktop
resolution using M_CURRENT_RESOLUTION DisplayFormat in MdispAlloc.
17- Mim and general processing functions
- New: MbufSetRegion function to set a region that defines the pixels of
interest for an image. The region can be stored either in a vector format
with the help of a graphics list, or in a raster format. Some processing
operations such as McodeRead, McodeVerify, MstrRead, McalFixture,
MblobCalculate... can use and take advantage of a region when set.
- New: MimFlatField function and context to perform an image flat field
correction operation.
- New: MimRearrange function and context to copy an entire list of ROIs from
a source buffer to a destination buffer at arbitrary locations.
- New: MimDeadPixelCorrection function and context to perform a dead pixel
correction operation.
- New: MimMatch function and context to compute the matching score between a
model image and a destination image using either the correlation or a
faster sum-of-differences based calculation.
- New: MimThin, M_BINARY3 processing mode which usually performs faster
than other thinning modes.
- New: MimPut() and MimGet() APIs to set / get values to Mim contexts.
- New: M_CUMULATIVE_VALUE histogram result (i.e. M_HIST_LIST result type) to
retrieve the cumulative histogram of an image.
- New: M_PERCENTAGE histogram combination flag result (i.e. M_HIST_LIST
result type) to be combined with M_VALUE or M_CUMULATIVE_VALUE to retrieve
the histogram information as percent values.
- New: M_PERCENTILE_VALUE histogram result (i.e. M_HIST_LIST result type) to
retrieve the percentile histogram values of an image.
- New: MimBinarize M_BIMODAL, M_PERCENTILE_VALUE, M_TRIANGLE_BISECTION_DARK,
and M_TRIANGLE_BISECTION_BRIGHT modes. Note that invoking the automatic
thresholding by setting the ConditionLow to M_DEFAULT is deprecated and
replaced by the M_BIMODAL flag.
- New: MimArith operations Square, Cube, Square root, Ln, Log10, Log2
and Exp_const are supported.
- New: MimMorphic M_LEVEL operation to perform a morphological window
leveling correction.
- New: MimMorphic M_OPEN and M_CLOSE morphological operations.
- New: MimMorphic M_TOP_HAT and M_BOTTOM_HAT morphological operations.
- New: MimStatMultiple to calculate pixel statistics on a sequence of
images.
- New: MimDraw to draw Mim context and results informations.
- New: MimGetResult2d to retrieve 2D results from a Mim result object.
- New: support overscan color for MimThin and MimThick in binary mode using
M_BINARY+M_OVERSCAN_REPLACE_MAX/M_OVERSCAN_REPLACE_MIN operation flag.
- New: M_RESULT_OUTPUT_UNITS to retrieve results such as M_POSITION_X and
M_POSITION_Y in world or pixel units (replaces M_OUTPUT_UNITS in
McalControl).
18- Interactive utilities
- New: GUI for the color module (matching context only).
- New: McolSave / Restore / and Stream interactive tools.
- New: Processing GUIs support RGB-to-HSL and HSL-to-RGB conversions
in the image menu.
- New: button in the Code Reader GUI to save a txt report containing most
of the results from a verify operation.
- Improvement: the metrology GUI pops a warning message when deleting a
feature with dependencies.
3. Fixed bugs
1- Code Reader module
- Fix: MicroPDF417 could fail to read with scan reverse enabled and the
target code rotated at 180 degrees.
- Fix: in the code verification related to a wrong interpretation of the
validation of the results in section J.1. Phase 1 of the ISO 24728 spec.
- Fix: in rare situations, a pair (CW + RAP) can "lose" against the score of
a pair (RAP + CW) in a MicroPDF417 code resulting in a bad reading result.
- Fix: string size min/max controls had no effect for Pharmacodes.
- Fix: large Maxicode symbols were often not read when using the adaptive
threshold (M_ADAPTIVE).
- Fix: M_SCAN_EDGE_DETERMINATION_GRADE was wrong for cross-row codes.
- Fix: in some rare cases, an GS1 Databar could have an erroneous decode
grade of F.
- Fix: McodeWrite did not scale codes optimally in the destination.
- Fix: M_SCAN_MODULATION result could be erroneous for several types of code
- Fix: the checksum calculation for code93 encoding 15 characters and more.
- Fix: McodeWrite/Read could crash when writing and reading "big" QR Code.
- Fix: McodeWrite for QR code was not using accurately the M_CELL_SIZE_MIN.
- Fix: ECMin verification fixed for 1D codes: now takes into account the
full
quiet zone as specified in ISO/IEC 15416.
- Fix: MicroPDF417 position in composite codes for the top left and bottom
rigth position values.
- Fix: possible floating point exception fixed only when changing the FPU
exception mode using MappAlloc.
2- Blob module
- Fix: MblobDraw was not multi-thread safe or exception safe with
M_DRAW_BLOBS_CONTOUR and M_DRAW_HOLES_CONTOUR options when chains
(M_CHAINS) were not previously calculated.
- Fix: when M_MAX_BLOBS condition was reached and when using the multi-core
capability, MblobCalculate could fail to properly merge the results.
- Fix: API protection: chain codes cannot be computed when SAVE_RUNS is set
to M_DISABLE.
- Fix: potential floaing point exception when explicitely casting a large
GetResult value using +M_TYPE_CHAR,...
- Fix: some timeout scenarios could cause a race condition leading to a crash.
3- Calibration
- Fix: in particular cases, McalDraw could generate wrong output with
calibrated destination buffer.
- Fix: the Tsai calibration process could lead to a floating-point exception
in extreme cases.
- Fix: McalInquire(M_RELATIVE_ORIGIN_X/Y/Z/ANGLE) on an image could return
incorrect values when the calibration was previously modified.
- Fix: McalTransformImage could produce wrong results for a specific
McalAssociate/McalRelativeOrigin\McalTransformImage sequence of calls.
- Fix: McalTransformImage on a child followed by McalTransformCoordinate
call could produce incorrect result values.
- Fix: the Tsai calibration could fail when using very large buffers.
- Fix: McalTransformImage with M_CORRECT_LENS_DISTORTION_ONLY could perform
a full correction, regarding the order of operations.
4- String Reader module
- Fix: a call to MstrRead or MstrDraw changed the color of the default
graphic context.
- Fix: a memory leak when restoring a fontless context.
- Fix: a potential floating point exception in x64 when reading with
a fontless context.
5- Edge Finder module
- Fix: MedgeDraw, M_DRAW_VALUE didn't work properly on calibrated images.
- Fix: MedgeSream on a result could temporarily change the M_OUTPUT_UNITS
of the calibration, when having calibrated results.
- Fix: MedgeGetResult(M_ELLIPSE_FIT_ANGLE) in pixels was not correct if
calculated on a calibrated image.
- Fix: streaming of M_NEIGHBOR_ANGLE and M_NEIGHBOR_ANGLE_TOLERANCE edge
results was not done properly.
6- Model Finder module
- Fix: MmodDefine was not working properly with a calibrated corrected image
- Fix: a potential overflow could lead to unstable results in rare cases.
- Fix: for synthetic models, the edges absent from the target image could
wrongly contribute to the model score.
- Fix: a memory leak in the preprocessing of a geometric controlled context
when the angle search was enabled.
- Fix: fractional M_EXTRACTION_SCALE was not properly handled for models
defined from an Edge result buffer.
- Fix: MmodGetResult(M_CHAIN_xxx + M_MODEL) crashed on remote systems.
- Fix: MmodPreprocess fix a case where the First level could be negative.
- Fix: possible wrong first/last level values when using calibrated
synthetic or CAD models.
- Fix: potential exception fixed in geometric controlled context with the
search angle range enabled and when the reference point happens to be
exactly on top of an edgel.
- Fix: MmodDraw M_DRAW_POSITION did not draw the reference point with
proper angle (model or result) when using calibrated model/target.
- Fix: MmodFind with M_GEOMETRIC_CONTROLLED on synthetic model was not
working properly when M_PIXEL_SCALE != 1.
- Fix: MmodFind "from seed" on calibrated image was not working properly
when M_FIRST_LEVEL != M_LAST_LEVEL.
- Fix: CAD or edge result model could have an innacurate calculated score.
7- Color module
- Fix: the sequence McolPreprocess, McolSetMethod, McolInquire could lead to
an abnormal behavior of the module for a few inquired controls.
- Fix: M_LABEL_AREA_IMAGE_SIZE_BIT, when no image generation is enabled,
must not be available in GetResult.
- Fix: McolAlloc with M_DEFAULT as object type was generating an exception.
- Fix: the average of the H band was not handling properly the angle wrap
around leading to wrong H values arounf 0 / 360 degrees.
8- Metrology module
- Fix: the angle of the output frame was not taken into account for the
angular measure of measured arcs and segments.
- Fix: retrieving result of a null edge data feature should not pop an error
message.
- Fix: MmetDraw(M_DRAW_TEMPLATE_REFERENCE) was not propagating the
calibration, if any, to destination image.
- Fix: MmetDraw(M_DRAW_TEMPLATE_REFERENCE) with LabelOrIndex M_GENERAL
reports an error.
- Fix: some MmetDraw did not consider the destination calibration of the
target image if any.
- Fix: in rare situations, a constructed intersection could lead to a crash
when no edgels were found.
9- 3dmap module
- Fix: M3dmapCalibrate could have an unpredictable behavior when the X axis
of the camera is perpendicular to the laser plane.
- Fix: M3dmapTriangulate documentation of parameters XPixelArrayPtr and
YPixelArrayPtr was incorrect.
10- Pattern Matching module
- Fix: MpatInquire with M_PROC_FIRST_LEVEL always returned 0 for a circular
overscan model instead of first level used.
- Fix: a model could not be found when the occurrence is at 180deg +-epsilon
and delta angle was set to 180deg.
- Fix: MinSpacing with normalized rotated models did not take into account
the effect of the rotation.
- Fix: occurrences at rotation partially occluded by the image border could
lead to unstable match results.
- Fix: occurrences reference point could fall outside a specified search
region when searching with rotation.
11- Measurement module
- Fix: MmeasStream + M_LOAD with two types of markers did not work.
- Fix: MmeasSetMarker + M_POSITION was not handled properly under.
specific conditions and sequence of Meas calls.
- Fix: M_POSITION_Y result values were wrong when using a calibrated
source buffer.
- Fix: MmeasGetResult(...M_BOX_CORNER_BOTTOM_LEFT, M_BOX_CORNER_...) results
were not converted into world coordinates.
- Fix: measuring with boxes with no rotation in a filpped host buffer
generated a crash.
- Fix: Angle result could have a +-180 degrees unstability.
- Fix: the contrast of a marker could be negative when using
M_SHEN/M_PREWITT filters.
- Fix: under some conditions, the angle of the box is returned instead of
the real angle of the edge/stripe when angle mode is enabled.
12- Registration module
- Fix: MregCalculate could fail due to the accumulation of
small numerical errors.
13- OCR module
- Fix: a potential exception in OCR General when reading in small regions.
14- Graphic functions
- Fix: a potential exception in MgraArc when one of the radii was set to 0.
- Fix: MgraControl(M_COLOR) and MgraControl(M_BACKCOLOR) were not
working properly.
15- Mim and general processing functions
- Fix: MimBinarize 32F->U8 was not a thread-safe operation.
- Fix: the default combination flag M_DRAW_PEAKS of the MimDraw operation
was not handled correctly.
- Fix: MthrControl(M_MP_XXX) with a specific MIL thread ID did not work.
- Fix: MimStat(...M_MIN, M_MAX...) did not return the correct results
in 64 bits.
- Fix: MimBinarize did not work as expected with a float source containing
values between 0 and 1 and a binary destination buffer.
- Fix: MbufCopyCond did not work correctly under x64 when the condition
buffer was an offsetted child of a binary buffer.
- Fix: MimPolarTransform could access memory outside the source image in
very rare situations.
- Fix: GPU buffers are not unlocked after some calls to Mim functions.
- Fix: MimBinarizeautomatic threshold on floating-point buffers did
not work as expected.
- Fix: potential fetch outside the source buffer for MimMorphic binary
operations.
16- Interactive utilities
- Fix: when opening a jpg image with the wrong extension in a GUI, and
without the appropriate license lead to a crash.
- Fix: when changing the current frame in a sequence, annotations were not
redrawn, even if the current selection would indicate otherwise.
- Fix: switching documents (images) in the GUIs with Ctrl+Tab or
Ctrl+Shift+Tab cleared the annotations.
- Fix: Model Finder GUI crashed when preforming a search in a small image
(less than 16x16) using a geometric controlled context.
- Fix: feature IDs in the Metrology GUI were inconsistent between the main
feature's tree and features in sub dialogs.
- Fix: modifying the model finder model reference point angle should require
to re-preprocess.
4. Mim functions and performance optimizations
- New: McolDistance and McolProject have been optimized using SIMD
instructions.
- New: McolDistance and McolProject have been optimized for multi-core (MP)
architectures.
- New: McolMatch with the vote method has been optimized using SIMD
instructions and for multi-core (MP) architectures.
- New: MpatFindModel has been optimized for multi-core (MP) architectures.
- New: MpatFindModel has been optimized using SIMD instructions for large
models.
- New: MblobCalculate has been optimized for multi-core (MP) architectures.
- New: MblobCalculate additional optimizations using SIMD instructions for
calculating the blob's RLE.
- New: MimMorphic, in binary mode, has been optimized using
SIMD instructions.
- New: MimResize M_AVERAGE with integer scale factors has been optimized
using SIMD instructions.
- New: MimResize M_AVERAGE has been optimized for multi-core (MP)
architectures.
- New: MimConvolve special moving average optimization for uniform
user-defined kernels.
- New: MimMorphic M_DILATE/M_ERODE special optimization for structural
elements filled with 0s.
- New: MimMorphic M_DILATE/M_ERODE special optimization for 1D structural
elements filled with 0.
- New: optimized MP predefined IIR filters (MimConvolve).
5. Board specifics
- Fix [ODC]: McodeStream(M_SAVE_REPORT) was not working properly on Odyssey
platforms.
- Fix [ODC]: a memory leak in MmodPreprocess on Odyssey when the kernel size
of the filter was large (using kernel mode and high filter smoothness).
6. Miscellaneous
- Predefined MIL contexts (%MIL_PATH%\..\Contexts\):
No new context.
- Calibration grid samples (%MIL_PATH%\..\images):
No new sample.
- Enhanced DispD3D DLL to display 3D depth maps.
- A Matrox M-Series graphics cards can now be used as a fingerprint for,
and storage of, a MIL license. Note that the Matrox display driver must be
started before the first call to MappAlloc.
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 10: Differences between MIL 9 Processing Pack 1 and MIL 9
Table of Contents for Section 10
1. Overview
2. New processing functionality and improvements
2.1. Color module (New!)
2.2. 3dmap module (New!)
2.3. Metrology module
2.4. Code Reader module
2.5. Edge Finder module
2.6. Calibration module
2.7. Blob module
2.8. String Reader module
2.9. Pattern Matching module
2.10. Model Finder module
2.11. Registration module
2.12. Mim and general processing functions
2.13. Interactive utilities
3. Fixed bugs
3.1. Metrology module
3.2. Code reader module
3.3. Blob module
3.4. String Reader module
3.5. Edge Finder module
3.6. Model Finder module
3.7. Ocr module
3.8. Pattern Matching module
3.9. Registration module
3.10. Mim and general functions
3.11. Interactive utilities
4. Mim functions and performance optimizations
5. Board specifics
6. Miscellaneous
-------------------------------------------------------------------------------
1. Overview
- MIL 9 Processing Pack 1 includes new processing functionality,
performance optimizations and general improvements.
- MIL 9 Processing Pack 1 also includes new ActiveMIL processing controls.
For more information, see the ActiveMIL readme file.
- MIL 9 is the minimum requirement for all upcoming Processing Packs until
the next major release.
2. New processing functionality and improvements
1- Color module (New!)
- New: a color module to perform color vision processing, including:
- supervised color matcher tools,
- color separation operators,
- and color distance operators.
The color tools can be parametrized using various controls.
Refer to the MIL Help file for more details.
2- 3dmap module (New!)
- New: a 3dmap module to generate 3D data, including:
- camera-laser scanning devices calibration tools
- camera-laser scanning devices depth map generation tools
- camera-laser scanning devices fill holes function
- multi-camera point triangulation tool.
3- Metrology module
- New: constructed feature construction methods (bisector between two lines,
intersection point between extended features,
- New: A local frame constructed feature can now be constructed from a
single point. The orientation of the local frame can be set using
MmetControl and is relative to the parent frame.
- Improvement: of the drawing of an infinite radial region.
4- Code Reader module
- New: code verifications for 1D codes (specifications: ISO 15416, 15417,
15420, 16388, 16390)
- New: code verifications for 2D codes (specifications: ISO 24724 (RSS),
15438 (PDF417 & Trunc. PDF417), 24723 (Composite), 24728 (MicroPDF417))
- New: M_APERTURE_MODE control flag to set the type of aperture to apply
when performing a code verification operation. When enabled, the aperture
mode can be set to an absolute size value or to a relative size to the
code module size.
- New: M_RELATIVE_APERTURE_FACTOR control flag to set the relative aperture
factor to the module size for optimal code verification operation.
- New: M_ABSOLUTE_APERTURE_SIZE control flag to set the absolute diameter
size for optimal code verification operation.
- New: M_MAXIMUM_CALIBRATED_REFLECTANCE to set the reference "maximum
calibrated value" used in the scan reflectance profiles analysis.
- New: M_MINIMUM_CALIBRATED_REFLECTANCE to set the reference "minimum
calibrated value" used in the scan reflectance profile analysis.
- New: M_LINEAR_COMPONENT combination flag to retieve results of the linear
component of a composite code. For example: McodeGetResult(..,
M_NUMBER_OF_ROWS + M_LINEAR_COMPONENT, ..)
- New: M_2D_COMPONENT combination flag to retieve results of the 2D
component of a composite code.
- New: M_NUMBER_OF_CODEWORDS to retrieve the number of codewords in the
symbol after a read or a verify operation for all 2D codes except for
non-ECC200 Data Matrix symbols.
- New: M_NUMBER_OF_ERROR_CORRECTION_CODEWORDS to retrieve the number of
error correction codewords in the symbol after a read or a verify
operation for all 2D codes except non-ECC200 Data Matrix symbols.
- New: M_NUMBER_OF_ERRORS to retrieve the number of errors found in the
symbol after a read or a verify operation for all 2D codes except for
non-ECC200 Data Matrix symbols.
- New: M_NUMBER_OF_ERASURES to retrieve the number of erasures found in the
symbol after a read or a verify operation for all 2D codes except for
non-ECC200 Data Matrix symbols.
- Updated: M_OVERALL_SYMBOL_GRADE now returns a numeric value instead of
a letter, in order to respect ISO 15415 and ISO 15416 specifications.
- New: M_NUMBER_OF_ROWS to retrieve the number of rows verified in the
symbol for 1D codes and RSS codes.
- New: M_NUMBER_OF_SCANS_PER_ROW to retrieve the number of scan reflectance
profiles for a given row for 1D codes and RSS codes.
- New: M_NUMBER_OF_SCANS to retrieve the total number of scan reflectance
profiles analyzed for 1D codes and RSS codes.
- New: M_ROW_OVERALL_GRADE to retrieve the overall grade per for 1D codes
and RSS codes.
- New: M_SCAN_DECODE_GRADE to retrieve per profile if the data was decoded
correctly.
- New: M_SCAN_SYMBOL_CONTRAST to retrieve per profile the symbol
contrast(SC).
- New: M_SCAN_SYMBOL_CONTRAST_GRADE to retrieve per profile the grade based
on the symbol contrast (SC) parameter.
- New: M_SCAN_REFLECTANCE_MINIMUM to retrieve per profile the minimum
reflectance (Rmin) parameter
- New: M_SCAN_REFLECTANCE_MINIMUM_GRADE to retrieve per profile the grade
based on the minimum & maximum reflectance (Rmin & Rmax) parameters.
- New: M_SCAN_REFLECTANCE_MAXIMUM to retrieve per profile the maximum
reflectance (Rmax) parameter.
- New: M_SCAN_EDGE_CONTRAST_MINIMUM to retrieve per profile the minimum edge
contrast (ECmin) parameter.
- New: M_SCAN_EDGE_CONTRAST_MINIMUM_GRADE to retrieve per profile the grade
based on the minimum edge contrast (ECmin) parameter.
- New: M_SCAN_MODULATION to retrieve per profile the modulation (MOD)
parameter.
- New: M_SCAN_MODULATION_GRADE to retrieve per profile the grade based on
the modulation (MOD) parameter.
- New: M_SCAN_ERN_MAXIMUM to retrieve per profile the element reflectance
non-uniformity maximum (ERNmax) parameter.
- New: M_SCAN_DEFECTS to retrieve per profile the defects parameter (based
on ERNmax and SC).
- New: M_SCAN_DEFECTS_GRADE to retrieve per profile the grade based on the
defects parameter.
- New: M_SCAN_DECODABILITY to retrieve per profile the decodability
parameter (V).
- New: M_SCAN_DECODABILITY_GRADE to retrieve per profile the grade based on
the decodability parameter (V).
- New: M_SCAN_REFLECTANCE_PROFILE_GRADE to retrieve per profile the scan
reflectance profile grade, which is equal to the lowest grade of any scan
profile grade.
- New: M_SCAN_PRINT_CONTRAST_SIGNAL to retrieve per profile the print
contrast signal (PCS) of the scan profile.
- New: M_SCAN_PROFILE_START_X to retrieve the x coordinate of the start of a
scan profile.
- New: M_SCAN_PROFILE_START_Y to retrieve the y coordinate of the start of a
scan profile.
- New: M_SCAN_PROFILE_END_X to retrieve the x coordinate of the end of a
scan profile.
- New: M_SCAN_PROFILE_END_Y to retrieve the y coordinate of the end of a
scan profile.
- New: M_SCAN_QUIET_ZONE to retrieve the worst quiet zone for Code 128,
EAN/UPC, Code 39, Interleaved 2 of 5, Codabar and Code 93.
- New: M_SCAN_QUIET_ZONE_GRADE to retrieve the grade based on the quiet zone
size for Code 128, EAN / UPC, Code 39, Interleaved 2 of 5, Codabar and
Code 93.
- New: M_SCAN_GUARD_PATTERN to retrieve the maximum interior guard pattern
element size for Rss codes, and composite codes containing a Rss code.
- New: M_SCAN_GUARD_PATTERN_GRADE to retrieve the grade based on the
interior guard pattern elements size for Rss codes, and composite codes
containing an Rss code.
- New: M_SCAN_WIDE_TO_NARROW_RATIO to retrieve the wide to narrow ratio for
Code 39, Interleaved 2 of 5 and Codabar.
- New: M_SCAN_WIDE_TO_NARROW_RATIO_GRADE to retrieved the grade based on the
wide to narrow ratio for Code 39, and Interleaved 2 of 5 and Codabar.
- New: M_SCAN_INTERCHARACTER_GAP to retrieve the maximum intercharacter gap
in the scan profile for Code 39 and Codabar.
- New: M_SCAN_INTERCHARACTER_GAP_GRADE to retrieve the grade based on the
maximum intercharacter gap for Code 39 and Codabar.
- New: M_CODEWORD_YIELD to retrieve the codeword yield result for the symbol
for MicroPDF417, PDF417 and Truncated PDF417.
- New: M_CODEWORD_YIELD_GRADE to retrieve the grade based on the codeword
yield for MicroPDF417, PDF417 and Truncated PDF417 codes.
- New: M_UNUSED_ERROR_CORRECTION_GRADE to retrieve the grade based on the
unused error correction is now supported for MicroPDF417, PDF417 and
Truncated PDF417 codes.
- Updated: M_START_STOP_PATTERN_GRADE to retrieve the grade based on the
analysis of the start/stop pattern (or RAPs) in the symbol for
MicroPDF417, PDF417 and Truncated PDF417 codes. Additionally, this result
is now a numeric value. See ISO 15415 for more details.
- Updated: M_DECODABILITY_GRADE to retrieve the grade based on the codeword
print quality assessment based on the decodability grade for MicroPDF417,
PDF417 and Truncated PDF417 codes only. For linear symbologies, see
M_SCAN_DECODABILITY_GRADE instead.
- Updated: M_DEFECTS_GRADE to retrieve the grade based on the codeword print
quality assessment based on the defects grade for MicroPDF417, PDF417 and
Truncated PDF417 codes only. For linear symbologies, see
M_SCAN_DEFECTS_GRADE instead.
- Updated: M_MODULATION_GRADE to retrieve the grade based on the codeword
print quality assessment based on the modulation grade for MicroPDF417,
PDF417 and Truncated PDF417 codes only. For linear symbologies, see
M_SCAN_MODULATION_GRADE instead.
- New: M_CODEWORD_DECODABILITY_GRADE to retrieve the decodability grade of
each codeword in the symbol.
- New: M_CODEWORD_DEFECTS_GRADE to retrieve the defects grade of each
codeword in the symbol.
- New: M_CODEWORD_MODULATION_GRADE to retrieve the modulation grade of each
codeword in the symbol.
- New:M_DRAW_SCAN_PROFILES drawing option to draw the scan profiles examined
after a verify operation.
- New: M_DRAW_REFLECTANCE_PROFILE drawing option to draw the scan
reflectance profiles analyzed by the verify operation.
- New: M_STRING_SIZE_MIN control flag to specify the minimum string length
requirement of a code read.
- New: M_STRING_SIZE_MAX control flag to specify the maximum string length
requirement of a code read.
- New: M_BEARER_BAR control flag to specify the presence of bearer bars
around the code in order to read at unknown search angles.
- New: Generation of a verification report can be done through McodeStream
with the operation M_SAVE_REPORT.
- New: BC412 now supports the M_ECC_CHECK_DIGIT error correction.
5- Edge Finder module
- New: M_FERET_ANGLE_SEARCH_START and M_FERET_ANGLE_SEARCH_END to set the
angle range limits used to compute the Feret min and max diameters and
angles.
6- Calibration module
- New: M_3D_ROBOTICS calibration context to fully calibrate the system
camera mounted on a robotic arm.
- New: M_ROBOT_BASE_COORDINATE_SYSTEM coordinate system to decribe the
robot base coordinate system (M_3D_ROBOTICS context only)
- New: M_NUMBER_OF_CALIBRATION_POSES, M_GLOBAL_AVERAGE_PIXEL_ERROR,
M_GLOBAL_AVERAGE_WORLD_ERROR,M_GLOBAL_MAXIMUM_PIXEL_ERROR, and
M_GLOBAL_MAXIMUM_WORLD_ERROR inquiries (M_3D_ROBOTICS context only).
- New: McalList and McalGrid M_ACCUMULATE mode to accumulate robot poses
(M_3D_ROBOTICS context only).
- New: McalInquireSingle to retrieve information about invidual poses
(M_3D_ROBOTICS context only).
- New: McalGrid supports a new type of calibration grid (M_CHESSBOARD_GRID).
- New: M_TRANSFORM_FILL_MODE McalControl flag to set the McalTransformImage
operation to the new modes M_FIT, M_CLIP, M_CORRECT_LENS_DISTORTION_ONLY,
or M_USER_DEFINED.
- New: M_TRANSFORM_IMAGE_WORLD_POS_X and M_TRANSFORM_IMAGE_PIXEL_SIZE_X
control flag to control the offset and scaling behavior of
McalTransformImage when setting the M_TRANSFORM_FILL_MODE to
M_USER_DEFINED.
- New: McalTransformImage M_EXTRACT_LUT_(X,Y) mode to retrieve the
transformation warping LUT.
- New: McalDraw function to draw various calibration results (extracted
grid points, calibration grid area, calibration grid filled area,...)
- New McalTransformCoordinate3DList to transform list of points between the
various 2D and 3D coordinate systems (M_TSAI_BASED and M_3D_ROBOTICS
contexts only).
- New: M_DISPLACE_RELATIVE_COORD McalList and McalGrid option. This mode can
be used to find the 3D posture of a known object (M_3D_ROBOTICS contexts
only)
- Deprecation: M_LOCATE_CAMERA_ONLY is renamed M_DISPLACE_CAMERA_COORD.
- New: the calibration module now supports Chessboard calibration grids.
7- Blob module
- New: the calculation operation stops when the max number of blob, set
using the M_MAX_BLOBS control value, is reached.
- New: M_MAX_BLOBS_END result value to retrieve if the M_MAX_BLOBS limit has
been reached.
- New: MblobGetRuns supports M_ALL to retrieve the runs for all blobs.
- New: M_INPUT_SELECT_UNITS flag in order to provide world units entries to
MblobSelect function.
- Improvement: improved robustness to lack of memory exceptions.
8- String Reader module
- New: fontless application specific contexts (ex. ANPR context) which do
not require the explicit definition of fonts. Refer to the MIL Help file
for more details.
- Improvement: improved robustness to lack of memory exceptions.
- Improvement: robustness has been improved when reading punctuations.
9- Pattern Matching module
- New: MpatStream function to stream pattern matching contexts.
- Improvement: Mpat model size limit has been removed.
10- Model Finder module
- New: GEOMETRIC_CONTROLLED contexts can now search within an angle range
setting the M_SEARCH_ANGLE_RANGE control value to M_ENABLE.
11- Registration module
- New: M_SUPER_RESOLUTION control to perform super-resolution image
composition.
- New: M_SR_PSF_TYPE, M_SR_PSF_RADIUS, and M_SR_SMOOTHNESS control flags to
setup the super-resolution operation.
- New: M_TRANSFORMATION_TYPE now supports M_TRANSLATION_ROTATION_SCALE.
12- Mim and general processing functions
- New: MimThin M_BINARY2 binary thinning method.
13- Interactive utilities
- New: support the subpixel drawing of the line approximation result in the
Edge Finder GUI.
3. Fixed bugs
1- Metrology module
- Fix: MmetGetResult could crash when retrieving result of non calculated
features.
- Fix: potential memory leak when releasing a context or or crash when
re-setting the reference template when the reference template is defined
from an edge result buffer.
- Fix: the status of a measured feature was failing is the coverage was
equal to the min coverage limit.
- Fix: when a feature is calculated using a non infinite region, then
changing to infinite region and re-calculating leads to invalid results.
- Fix: an angularity tolerance with a range 0-360 was always returning a
fail status intead of a pass status.
- Deprecation: MmetControl M_INDEX selection flag is renamed M_OCCURRENCE.
- Fix: the point construct using the max distance method between a circle or
an arc and another feature was sometimes returning two points instead of
one.
- Fix: when using calibration, accuracy of edgel orientation has been
improved.
- Fix: MmetCalculate could crash when a NULL context ID was provided.
- Fix: MmetCalculate could crash when a NULL system ID was provided.
- Fix: inner/outer fit segment was not always returning the expected result.
2- Code Reader module
- Fix: RSS codes decoding was expecting quiet zones, but only the guard
patterns are required.
- Fix: the timeout end result did not return the appropriate
M_STATUS_TIMEOUT_END status.
- Fix in some rare situation for 1D codes, the reading operation could fail
when setting the angle range to 180 degrees.
- Fix: some RSS Expanded codes could be decoded as RSS Expanded Stacked
codes.
- Fix: CC-A MicroPDF417 symbol could fail decoding some codewords.
- Fix: for 1D codes, the position result could be false when the code angle
was near a multiple of 90 degrees.
- Fix: UCC-EAN128 /PDF417 Composite codes generated by McodeWrite was not
always having the mandatory quiet zones size.
- Fix: in rare situations, PDF417 error correction mechanism could lead to
false read.
- Fix: writing a composite code with a very long string could lead to a crash.
- Fix: Pharmacode position could be offsetted when scan reverse was enabled
or when reading multiple models.
- Fix: MicroPDF417 decoding has been improved when the microPDF417 is part
of a composite code.
- Fix: the cell number X was incorrect for Truncated PDF417 symbols.
- Fix: reading a code at very low speed or with multiple code models could
enable the scan reverse, wrongly overriding the user value.
- Fix: the maximum cell size limitation for 2D matrix code (DM, QR and
Maxicode) has been removed.
- Fix: general bug fixes have improved the reading of composite code
containing a RSS code, Maxicode, PDF417, Truncated PDF417, composite codes
with CC-C, Data Matrix ECC200.
- Fix: general bug fixes have improved the verification of MicroPDF417, and
composite codes with CC-C.
3- Blob module
- Fix: potential MblobSelect crash when merging blobs with a selection
criterion using the blob area feature.
- Fix: MblobControl(M_TIMEOUT, M_DEFAULT) was not deactivating the a
previously set timeout.
4- String Reader module
- Fix: the homogeneity score was not always retruning 100% for a string
composed of a single character.
- Fix: the drawing of the underscore character was not respecting the
baseline value.
- Fix: some string experts error reports were ambiguous and have been
clarified.
5- Edge Finder module
- Fix: MedgeGetNeighbors was not working properly when the
M_SEARCH_RADIUS_MAX control was not previously set.
- Fix: MedgeGetNeighbors has been improved for large search radius values.
- Fix: the follwing combination [crest context, filter type Shen, filter
mode kernel], the calculation of the edges could generate a crash.
- Fix: the drawing of the chain approximation segments was erroneous under
Windows Vista 64 bits.
6- Model Finder module
- Fix: a potential crash when matching in an edge result buffer calculated
using an M_CREST edge context type.
- Fix: potential crash durng a match when the M_DETAIL_LEVEL control is set
to M_LOW.
- Fix: the search region was erroneous when setting a pixel scale !=1 for a
DXF based CAD model.
- Fix: small radius arcs from DXF based models were not properly extracted.
7- OCR module
- Fix: potential crash of a read operation when using an M_GENERAL OCR
context type with the M_TARGET_CHAR_SPACING set to M_SAME or to a user
defined amount of spaces.
- Fix: in very rare situations, two successive readings could lead to
different results.
8- Pattern Matching module
- Fix: parameter checking fixes for MpatControl and MpatInquire functions.
9- Registration module
- Fix: potential memory leak in MregSave/MregRestore/MregStream of a result
element.
10- Mim and general processing functions
- Fix: in rare situation, a potential crash in MimWarp when using M_WARP_LUT
mode has been fixed.
- Fix: MimEdgeDetect accuracy has been improved when using float buffers.
- Fix: small verticals Mpat models (size X < 4) are now supported.
- Improvement: performance of multi-thread MIL applications is improved when
the processing involves calls to modules (blob, ocr,...)
11- Interactive utilities
- Fix: when multiple context were opened, adding a feature in one context
was adding the feature in all context.
- Fix: context calibration information was lost when cancelling the feature
properties dialog.
4. Mim functions and performance optimizations
- Improvement: MimLocatePeak1D, MimLocateEvent, and MimStat reoptimization
and general optimizations of the image processing functions.
- New: MIL now supports multiprocessing optimizations for multi-core and
multi-processor machines.
- New: M_MP_USE flag to enable / disable the use of MP capability either
globally using MappControl() or the at thread level using MthrControl().
- New: M_MP_MAX_CORES_PER_THREAD flag to set globally a maximum number of
cores to use in any processing thread using MappControl()
- New: M_MP_MAX_CORES flag to set at the thread level a maximum number of
cores to use using MthrControl()
- New: M_MP_CORES_NUM flag to retrieve the number of cores available to the
application using MappInquire()
- New: M_MP_MAX_CORES_EFFECTIVE to retrieve the effective number of cores
used for MP processing by the given thread using MthrInquire()
- New: the following image processing functions have been optimized for
multiprocessing usage (MP):
- MbufBayer
- MbufClear
- MbufCopy
- MbufCopyClip
- MbufCopyColor2d
- MbufCopyColor
- MbufCopyCond
- MbufCopyMask
- MimArith
- MimArithMultiple
- MimBinarize
- MimClip
- MimConnectMap
- MimConvert
- MimConvolve (except IIR predefined filters)
- MimDilate/MimErode
- MimEdgeDetect
- MimFindExtreme
- MimHistogram
- MimLutMap
- MimMorphic (except AreaOpen and AreaClose operations)
- MimOpen/MimClose
- MimProject
- MimRank
- MimResize
- MimRotate
- MimShift
- MimThick
- MimThin
- MimTransform
- MimWarp
- MimFlip
- MimLocateEvent
- MimLocatePeak1d
- MimStat
- MimTranslate
- MimDeinterlace
- MimPolarTransform
- JPEG and JPEG2000 Compression
5. Board specifics
- Fix [ODC]: The system hangs when a wrong template id is passed to a
Metrology context allocated on an odyssey system.
6. Miscellaneous
- Predefined contexts
The new %MIL_PATH%\..\Contexts\ directory includes the following predefined
MIL contexts:
+ Ocr context: SEMI_M12-92.mfo
+ Ocr context: SEMI_M13-88.mfo
+ Ocr context: SEMI.mfo
+ Ocr context: MICR_CMC_7.mfo
+ Ocr context: MICR_E_13B.mfo
+ Ocr context: OCR_A_BT.mfo
+ Ocr context: OCR_B_BT.mfo
+ Code context: SEMI_T1-95r0303.mco
+ Code context: SEMI_T2-0298E.mco
+ Code context: SEMI_T7-0303.mco
+ String Reader context: FONTLESS_ANPR.msr
+ String Reader context: FONTLESS_EUROPEAN_ANPR.msr
+ String Reader context: FONTLESS_MACHINE_PRINT.msr
- Calibration grid samples
Calibration grid samples are included in the %MIL_PATH%\..\images directory:
+ CircleCalibrationGrid_15x16_Letter.pdf
+ CircleCalibrationGrid_18x22_Letter.pdf
+ ChessboardCalibrationGrid_15x16_Letter.pdf
+ ChessboardCalibrationGrid_18x22_Letter.pdf
-------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Section 11: Differences between MIL 9 and MIL 8 Processing Pack 4
Table of Contents for Section 11
1. Overview
2. New processing functionality and improvements
2.1. Metrology module
2.2. Code Reader module
2.3. Calibration module
2.4. Blob module
2.5. String Reader module
2.6. Measurement module
2.7. Registration module
2.8. Pattern Matching module
2.9. Mim and general processing functions
2.10. Interactive utilities
3. Fixed bugs
3.1. Metrology module
3.2. Code Reader module
3.3. Calibration module
3.4. Blob module
3.5. String Reader module
3.6. Measurement module
3.7. Edge Finder module
3.8. Registration module
3.9. Model Finder module
3.10. OCR module
3.11. Pattern Matching module
3.12. Mim and general functions
3.13. Interactive utilities
4. Mim functions and performance optimizations
5. Board specifics
-------------------------------------------------------------------------------
1. Overview
- MIL 9 includes new processing functionality, performance optimizations
and general improvements.
- MIL 9 also includes new ActiveMIL processing controls. For more
information, see the ActiveMIL readme file.
- MIL 9 is the minimum requirement for all upcoming Processing Packs until
the next major release.
2. New processing functionality and improvements
1- Metrology module
- New: the diameter of a circle feature can be retrieved.
- New: the circle radius and arc radius tolerances can be defined.
- Improvement: the MmetAddFeature BuildOperation parameter can now
be inquired.
- Improvement: the MmetAddFeature FeatureLabelArray parameter can now
be inquired.
2- Code Reader module
- New: the numerical values of the parameters estimated during a call to
McodeVerify can be retrieved.
- New: support for M_FOREGROUND_ANY for linear 1D codes and microPDF417.
- Improvement: high accuracy support for position results when using
adaptive threshold.
- Improvement: the accuracy of geometric results (position, length,
height, width, cell size) has been improved.
- Improvement: general performance improvements in decoding and more
specifically, for QR codes.
- Improvement: the estimation of the position of RSS codes has been
improved.
- Improvement: reading composite codes at an angle is more robust
- Improvement: reading RSS Expanded Stacked has been improved when at
low speed.
- Improvement: the position of the corners of the read code can
be retrieved.
3- Calibration module
- New: M_GRID_CORNER_HINT_X/Y flag which allows for a user defined
position to help solve the grid orientation ambiguity when the grid
is reverted or heavily rotated (ex. 45 degrees).
- New: M_CALIBRATION_STATUS to retrieve information regarding the failure
of a calibration.
- New: M_Y_AXIS_UP to inquire about the orientation of the Y-axis.
- New: M_TSAI_BASED calibration method for full 3D camera positioning
with the estimation of the extrinsic and intrinsic parameters of the
camera.
- New: M_CALIBRATION_PLANE control to set the associated calibration grid
coordinate system (M_TSAI_BASED method only).
- New: M_DISTORTION_RADIAL_1 to retrieve the estimated lens radial
distortion (M_TSAI_BASED method only).
- New: M_FOCAL_LENGTH to retrieve the estimated effective pinhole focal
length in horizontal pixel units (M_TSAI_BASED method only).
- New: M_CCD_ASPECT_RATIO to set the aspect ratio of the CCD
(M_TSAI_BASED method only).
- New: M_PRINCIPAL_POINT_X/Y to set the position of the optical center
(M_TSAI_BASED method only).
- New: M_LINK_TOOL_AND_CAMERA to set a rigid link between the tool
coordinate system and the camera coordinate system (M_TSAI_BASED method
only).
- New: M_LOCATE_CAMERA_ONLY McalList/McalGrid flag to re-calibrate the
camera position relative to a new grid system (M_TSAI_BASED method
only).
- New: M_TOOL_POSITION_X/Y/Z to translate the tool coordinate system
relative to the absolute coordinate system.
(note: in MIL 9 Processing Pack 1, the name has changed to
M_DISPLACE_CAMERA_COORD)
- New: McalSetCoordinateSystem() to move the position of a coordinate
system relative to another coordinate system using various
transformation methods (M_TSAI_BASED method only).
- New: McalGetCoordinateSystem() to inquire the position of a coordinate
system relative to another coordinate system (M_TSAI_BASED method
only).
- New: M_CALIBRATION_CHILD_OFFSET_X/Y to specify the child offsets when
associating a calibration to a buffer or to a child buffer.
4- Blob module
- New: M_LABELED_TOUCHING flag which allows the detection of blobs from
labels, where touching blobs with different labels are considered to be
distinct blobs.
5- String Reader module
- New: the bounding box of the string can be retrieved.
- New: the bounding box of the individual characters can be retrieved.
- Improvement: a label can be associated to a string model.
- Improvement: a label can be associated to a font model.
6- Measurement module
- New: the measurement regions can be sub-divided into sub regions for a
better fit of the position and the angle of the edge.
7- Registration module
- New: a score of overlapped regions of two registered images can be
enabled and retrieved.
- New: support for alpha-blending of the overlapped regions.
8- Pattern Matching module
- New: M_SAVE_SUMS control for the MpatSetSearchParameter() function to
enable the saving of results of the individual sums involved in the
correlation formula.
- New: additional flags for MpatGetResult to retrieve the results of the
individual sums involved in the correlation formula (M_SUM_I, M_SUM_II,
M_SUM_IM, M_SUM_M, M_SUM_MM).
- New: additional flags for MpatGetResult to retrieve the number of
pixels used to compute the correlation (M_NUMBER_OF_PIXELS).
9- Mim and general processing functions
- New: MimRestore, MimSave and MimStream functions to save and restore
Mim contexts.
- New: MimConvert matrix multiplication for user defined color
conversions.
- New: added support for Magnitude and Phase in MimTransform reverse FFT
operation.
- New: MimTransform new M_POLAR method to transform cartesian system
values from/to polar system values.
- New: new 2x2 average Bayer demosaicing method.
- New: ability to stop a focus operation for MdigFocus.
- Improvement: MbufBayer(M_ADAPTIVE, M_COLOR_CORRECTION) now supports
16 bits buffers.
- Improvement: MimTransform(M_FFT, M_REVERSE) with floating-point sources
and integer destinations accepts the M_SATURATION flag.
10- Interactive utilities
- New: TSAI based calibration mode has been added to the calibration
dialog.
- New: the Metrology module now supports interactive dialogs.
3. Fixed bugs
1- Metrology module
- Fix: the drawing of labels for line feature, edgel feature and
parallelism tolerance was erroneous when calibration was used.
- Fix: a geometric ambiguity could have lead to an erroneous distance
tolerance status.
- Fix: the results of edgel features were not flushed when no new edgel
features were found in the following image.
- Fix: the result of the coverage of a circle was not updated to the
measured value.
- Fix: the result of the general status was not updated when the context
did not contain any tolerance definitions.
- Fix: a geometric construction could lead to an ambiguity, resulting in
erroneous min/max distance tolerance calculations.
- Fix: inner fit and outer fit of segment features have been improved
when using metrology with calibration.
- Fix: the feature status was M_FAIL when the measured coverage value was
equal to the min coverage value of the tolerance range.
- Fix: the maximum distance between a circle/arc and another feature
could have lead to a non-unique solution.
- Fix: MmetStream(M_SAVE/M_LOAD/M_INQUIRE_SIZE_BYTE) lead to a failure
when using an M_NULL system ID.
- Fix: MmetGetResult applied to a feature that was not found lead to
a fatal failure.
- Fix: MmetCalculate applied to an M_NULL context ID lead to a fatal
failure.
- Fix: an MmetGetResult call with invalid parameters gave invalid error
messages.
2- Code Reader module
- Fix: when a timeout was reached, the status of a Data Matrix read
operation was erroneous.
- Fix: the position estimation of MicroPDF417 codes was erroneous.
- Fix: the position estimation of Maxicode codes with the adaptive
threshold method was erroneous.
- Fix: the threshold value was not updated when reading a maxicode with
a user defined threshold value.
- Fix: reading a code at 90 degrees with adaptive threshold was not
working properly.
- Fix: the high accuracy position of an RSS code was not valid when the
angle was outside the range of [0,90].
- Fix: the returned position for M_PDF417 and M_TRUNCATED_PDF417 code
types was always 0.
- Fix: there was an offset in the position after a successful read of
2D code types when the presearch was enabled.
- Fix: the result of the cell size has been improved when reading
discrete linear codes.
- Fix: the comparison of the user supplied string in McodeVerify has
been fixed.
- Fix: the M_ECC_CORRECTED_NUMBER result was erroneous.
- Fix: McodeInquire of the default value of the M_POSITION_ACCURACY
control returned M_DEFAULT instead of M_LOW.
- Fix: when reading a code with angle detection and high position
accuracy mode, the position result was not calculated properly in rare
situations.
- Fix: McodeVerify of matrix codes (Data Matrix & QRCode) had a bad Unused
ECC grades value in rare situations.
- Fix: McodeRead and McodeInquire on a same code context could not be
used safely in different threads.
3- Calibration module
- Fix: a floating-point overflow could have occurred when an
M_INVALID_POINT was present.
- Deprecation: M_OUTPUT_COORDINATE_SYSTEM has been renamed to
M_OUTPUT_UNITS.
- Deprecation: M_CAMERA_POSITION_X/Y/Z controls are deprecated.
- Documentation: The former camera coordinate system has been replaced by
tool coordinate system. The new camera coordinate system now refers to
the coordinate system attached to the pinhole of the camera.
4- Blob module
- Fix: multiple calls to MblobCalculate with M_BLOB_IDENTIFICATION set to
M_LABELED could have lead to erroneous blob results.
- Fix: the accuracy of blob features has been improved for large buffers.
- Fix: the results of contact points retrieved using MblobGetResult were
inconsistent with the results retrieved from MblobGetResultSingle.
5- String Reader module
- Fix: internal fixes were applied to improve the reading of accentuated
characters.
- Fix: in rare situations the module could have generated an invalid
floating point operation.
6- Measurement module
- Fix: the edge strength result was erroneous.
- Fix: certain edges with very low contrast were ignored when they should
not have been.
- Fix: edge extremity results will now be returned in an order consistent
with the measurement box orientation.
7- Edge Finder module
- Fix: in rare situations, computation of the equivalent IIR filter could
have lead to an invalid memory access.
- Fix: a potential division by 0 in the ellipse fit has been fixed.
- Fix: MedgeGetNeighbors with M_SEARCH_RADIUS_MAX control set to
M_INFINITE was not properly interpreted.
- Fix: in rare situations, MedgeCalculate could be slowed down by
inefficient memory access.
8- Registration module
- Fix: in some situations an overflow could have happened which affected
the robustness of MregCalculate.
- Fix: a potential memory leak in MregCalculate has been fixed.
- Fix: failure to register one image prevented the registration of other
images during a call to MregCalculate.
9- Model Finder module
- Fix: M_SEARCH_OFFSET_X/Y control was not properly updated when set to
M_INFINITE.
- Fix: in extremely rare situations, MmodFind could have lead to an
infinite loop.
- Fix: the weighted mask was not correctly supported when using
M_GEOMETRIC_CONTROLLED.
10- OCR module
- Fix: MocrInquire did not properly support the M_TYPE_MIL_ID flag.
- Fix: a parameter verification issue when running in unicode has been
fixed in the MocrVerifyString function.
11- Pattern Matching module
- Fix: MpatAllocResult with a large number of entries lead to a fatal
failure.
- Fix: in rare situations the score result was lower than expected.
- Fix: MpatAllocAutoModel of small models lead to a fatal failure.
12- Mim and general processing functions
- Fix: the B component of BGR32 was lost in the MimArith(ADD) operation
when the source and destination were in different types of memory (Host
and Video).
- Fix: the accuracy of MimProject(M_90_DEGREE) has been improved for
large 8-bits buffers.
- Fix: an internal rounding of the user-supplied MinIntensity threshold
value in MimLocatePeak1d has been corrected.
- Fix: MimFree lead to a crash when called with an invalid id.
- Fix: MimAllocResult() with the M_DEFAULT flag as the second parameter
caused a "Not enough host memory to allocate buffer" error.
- Fix: MbufExportSequence was not working properly when dealing with
images of different sizes.
- Fix: MimConvolve with a separable kernel and transparent overscan has
been improved.
- Fix: MbufBayer(M_ADAPTIVE) saturation has been fixed.
- Fix: MimArith(M_NOT) has been fixed for small buffers whose size x
equals 8.
13- Interactive utilities
- Fix [All]: under some conditions, the selection of results was not
working properly.
- Fix [Edge]: the update of drawing annotations was sometimes done in
the wrong display.
- Fix [Metrology]: drawing annotations were not updated properly when a
sub-item was selected.
- Fix [Metrology]: controls for missing features have been added.
- Fix [Metrology]: an incorrect behavior which occurred when drawing and
zooming the active edgels in the template display buffer has been fixed
- Fix [Metrology]: the addition of a new feature or tolerance was applied
to all open contexts.
- Fix [Metrology]: the behavior of the scrollbars for the template
display buffer has been improved.
- Fix [StringReader]: the “Operations” group was not updated properly
4. Mim functions and performance optimizations
- General optimizations of the image processing functions.
5. Board specifics
- New [GPU]: GPU accelerations of image processing operations
(For more information, see the GPU readme file).
- New [Helios]: MbufBayer with M_AVERAGE_2X2 has been optimized for
the Matrox Helios.
- New [Helios]: MimLocatePeak1d has been optimized for the Matrox Helios.
- Fix [Helios]: a potential memory leak has been fixed when using the blob
module for the Matrox Helios system.
- New [Helios]: MimArithMultiple(M_OFFSET_GAIN) is accelerated by the
Matrox Helios for 8 and 16 bits buffers.
- New [Odyssey]: MbufBayer with M_AVERAGE_2X2 has been optimized for
the Matrox Odyssey.
- New [Odyssey]: MimLocatePeak1d has been optimized for the Matrox Odyssey.
- New [Odyssey]: MimConvert with matrix conversion has been optimized
for the Matrox Odyssey.
- New [Odyssey]: MimArithMultiple(M_OFFSET_GAIN) is accelerated by the
Matrox Odyssey for 8 and 16 bits buffers.
- Fix [Odyssey]: MimResize has been improved for the Matrox Odyssey.
- Fix [Odyssey]: calibration propagation mechanism has been fixed for
the Matrox Odyssey.