MimAugment

See also

User Guide:

Advanced techniques Augmentation

Reference:

MbufAlloc1d MbufAllocColor MbufInquire MimAlloc MimAllocResult MimControl MimInquire

Availability

Not available in MIL-Lite

Available in MIL

Available on Windows

Available on Linux

Not supported on:

Not supported on Matrox Indio (Update 53 or Update 68) (w)

Not supported on Matrox Concord PoE with ToE (Update 82) (z)

Fully supported on:

Fully supported on Host system (a)

Fully supported on Matrox GigE Vision driver (MIL 10, Update 25, Update 54, or Update 85)* (c)

Fully supported on Matrox Morphis (g)

Fully supported on Matrox Morphis QxT (h)

Fully supported on Matrox Orion HD (i)

Fully supported on Matrox Radient eCL (j)

Fully supported on Matrox Radient eV-CXP (MIL 10 or Update 29)* (k)

Fully supported on Matrox Solios eA/XA (l)

Fully supported on Matrox Solios ecl/xcl/ev-cl (m)

Fully supported on Matrox USB3 Vision driver (Update 19) (o)

Fully supported on Matrox RadientPro-CL (Update 7) (p)

Fully supported on Matrox Radient eV-CL (Update 27 or 74) (r)

Fully supported on Matrox Iris GTR (Update 28 or Update 68) (t)

Fully supported on Matrox Clarity UHD (Update 36 or 76) (u)

Fully supported on Matrox GenTL driver (Update 34) (v)

Fully supported on Matrox Rapixo CXP (Update 75 or Update 84) (y)

Fully supported on Matrox Rapixo CL Pro (Update 96) (aa)

* Refer to the MIL system-specific information
in the MIL Reference for the minimum update required.

For any information regarding a MIL system added during a MIL Update, see the MIL system’s release note

Available on Non-Matrox computer

Available on Matrox 4Sight-X

Available on Matrox 4Sight GP/GPm

Available on Matrox Supersight

Function map

Examples

adaptivefiltering.cpp
classcnncompletetrain.cpp
classtreeensembletrain.cpp
mimaugment.cpp

Function Map

- AugmentationContextImId

- SrcImageBufId

- DstImageBufOrAugmentationResultImId

- SeedValue

Table:

For specifying the seed that will generate the randomized settings

- ControlFlag

Table:

For specifying the operation to perform

Compilation information

Synopsis

Perform an augmentation operation on an image.

Syntax

void MimAugment(

MIL_ID AugmentationContextImId,	//in
MIL_ID SrcImageBufId,	//in
MIL_ID DstImageBufOrAugmentationResultImId,	//in
MIL_INT64 SeedValue,	//in
MIL_INT64 ControlFlag	//in

)

Description

This function performs an augmentation operation that allows you to create a plausible variation of an image. This is done by applying a specified ordered set of image processing operations with randomized settings within a specified range. To establish the image processing operations that MimAugment() can perform, call MimControl() with M_AUG_..._OP.

MimAugment() places its results in the specified destination buffer, which can be an image or an augmentation result.

The image processing operations that you specify MimAugment() to perform (M_AUG_..._OP) are prioritized as follows:

1. Affine operations, such as: M_AUG_ROTATION_OP M_AUG_SCALE_OP M_AUG_TRANSLATION_X_OP M_AUG_TRANSLATION_Y_OP M_AUG_ASPECT_RATIO_OP M_AUG_SHEAR_X_OP M_AUG_SHEAR_Y_OP	2. Structure operations, such as: M_AUG_DILATION_OP M_AUG_EROSION_OP M_AUG_DILATION_ASYM_OP M_AUG_EROSION_ASYM_OP	3. Geometric operations, such as: M_AUG_FLIP_OP M_AUG_CROP_OP
4. Intensity operations, such as: M_AUG_INTENSITY_MULTIPLY_OP M_AUG_INTENSITY_ADD_OP M_AUG_GAMMA_OP M_AUG_LIGHTING_DIRECTIONAL_OP M_AUG_SATURATION_GAIN_OP M_AUG_HUE_OFFSET_OP M_AUG_HSV_VALUE_GAIN_OP	5. Linear filter operations, such as: M_AUG_SMOOTH_DERICHE_OP M_AUG_SMOOTH_GAUSSIAN_OP M_AUG_BLUR_MOTION_OP M_AUG_SHARPEN_DERICHE_OP	6. Noise operations, such as: M_AUG_NOISE_GAUSSIAN_ADDITIVE_OP M_AUG_NOISE_MULTIPLICATIVE_OP M_AUG_NOISE_SALT_PEPPER_OP

To modify the order in which MimAugment() prioritizes the specified image processing operations, call MimControl() with M_PRIORITY.

Note that this function reference has not been updated for a MIL system added during a MIL update. Refer to the MIL system's release note to see which MIL system’s documentation you should use in its place and any possible differences.

Parameters

This function is not supported on the selected boards.

This function reference has not been updated for the selected MIL system. To show the content of this page, choose a second MIL system; refer to the MIL system's release note to see which MIL system’s documentation to choose and any possible differences.

Parameters

AugmentationContextImId

Specifies the identifier of the augmentation context. The context must have been allocated using MimAlloc() with M_AUGMENTATION_CONTEXT.

SrcImageBufId

Specifies the identifier of the source image buffer. If you are preprocessing the augmentation context (M_PREPROCESS), set this parameter to M_NULL.

The image buffer can be grayscale or color (1- or 3-bands); it can be a 1-, 8-, or 16-bit unsigned buffer, or a 32-bit floating-point buffer (with values that are normalized between 0.0 and 1.0). You must allocate this buffer using MbufAlloc...() with M_IMAGE + M_PROC.

You can perform any image processing operation (M_AUG_..._OP) on all supported source image buffers, except 1-bit unsigned images or 1-band images.

1-bit unsigned images do not support the following:

1-band images do not support the following:

When specifying a color (3-band) image, MimAugment() assumes an RGB color space by default. Operations that require a different color space (for example, HSV) internally convert the source image color. If your source image is not RGB, the conversion produces erroneous results.

DstImageBufOrAugmentationResultImId

Specifies the identifier of the destination in which to write the result of the augmentation transformation. The identifier must be for an image buffer or an augmentation image processing result buffer. If you are preprocessing the augmentation context (M_PREPROCESS), set this parameter to M_NULL.

The image buffer can be grayscale or color (1- or 3-bands); it can be a 1-, 8-, or 16-bit unsigned buffer, or a 32-bit floating-point buffer. You must allocate this buffer using MbufAlloc...() with M_IMAGE + M_PROC.

The augmentation result buffer must have been allocated on the required system using MimAllocResult() with M_AUGMENTATION_RESULT.

If you specify an image buffer that has an ROI associated with it, the ROI must be in raster format; otherwise, you will get an error.

SeedValue

Specifies the initial value that will generate the randomized settings for the image processing operations used to perform the augmentation. This parameter must be set to one of the following values.

For specifying the seed that will generate the randomized settings
Value	Description
M_DEFAULT	Specifies the default value. This depends on the seed mode, which is set using MimControl() with M_AUG_SEED_MODE. To use M_DEFAULT, the seed mode must be set to M_RNG_AUTO (default) or M_RNG_INIT_VALUE. (summarize) Specifies the default value. (more details...)
M_NULL	Specifies that you are preprocessing the augmentation context (M_PREPROCESS).
Value >= 0	Specifies a user-defined seed value. You can only specify a user-defined seed if MimControl() with M_AUG_SEED_MODE is set to M_USER_DEFINED_SEED. You cannot specify a value that is greater than the maximum value of a MIL_INT32. (summarize) Specifies a user-defined seed value. (more details...)

ControlFlag

Specifies whether MimAugment() should perform the augmentation, or should explicitly preprocess the augmentation context. If you do not explicitly preprocess the context (M_PREPROCESS), MimAugment() preprocesses it the first time you call the function.

This parameter must be set to one of the following values.

For specifying the operation to perform
Value	Description
M_DEFAULT	Performs the augmentation.
M_PREPROCESS	Preprocesses the specified augmentation context explicitly. In this case, you must set the SrcImageBufId, DstImageBufOrAugmentationResultImId, and SeedValue parameters to M_NULL. (summarize) Preprocesses the specified augmentation context explicitly. (more details...)

Compilation information

Header	Include mil.h.
Library	Use mil.lib; milim.lib.
DLL	Requires mil.dll; milim.dll.