Image Processing

Modules
	Color space processing
	Histogram Calculation
	Structural Analysis and Shape Descriptors
	Hough Transform
	Feature Detection

Detailed Description

Classes
class	cv::cuda::CannyEdgeDetector
	Base class for Canny Edge Detector. : More...
class	cv::cuda::TemplateMatching
	Base class for Template Matching. : More...

Enumerations
enum	cv::cuda::ConnectedComponentsAlgorithmsTypes {   cv::cuda::CCL_DEFAULT = -1 ,   cv::cuda::CCL_BKE = 0 }
	Connected Components Algorithm. More...

Functions
void	cv::cuda::bilateralFilter (InputArray src, OutputArray dst, int kernel_size, float sigma_color, float sigma_spatial, int borderMode=BORDER_DEFAULT, Stream &stream=Stream::Null())
	Performs bilateral filtering of passed image.
void	cv::cuda::blendLinear (InputArray img1, InputArray img2, InputArray weights1, InputArray weights2, OutputArray result, Stream &stream=Stream::Null())
	Performs linear blending of two images.
void	cv::cuda::connectedComponents (InputArray image, OutputArray labels, int connectivity, int ltype, cv::cuda::ConnectedComponentsAlgorithmsTypes ccltype)
	Computes the Connected Components Labeled image of a binary image.
void	cv::cuda::connectedComponents (InputArray image, OutputArray labels, int connectivity=8, int ltype=CV_32S)
Ptr< CannyEdgeDetector >	cv::cuda::createCannyEdgeDetector (double low_thresh, double high_thresh, int apperture_size=3, bool L2gradient=false)
	Creates implementation for cuda::CannyEdgeDetector .
Ptr< TemplateMatching >	cv::cuda::createTemplateMatching (int srcType, int method, Size user_block_size=Size())
	Creates implementation for cuda::TemplateMatching .
void	cv::cuda::meanShiftFiltering (InputArray src, OutputArray dst, int sp, int sr, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 5, 1), Stream &stream=Stream::Null())
	Performs mean-shift filtering for each point of the source image.
void	cv::cuda::meanShiftProc (InputArray src, OutputArray dstr, OutputArray dstsp, int sp, int sr, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 5, 1), Stream &stream=Stream::Null())
	Performs a mean-shift procedure and stores information about processed points (their colors and positions) in two images.
void	cv::cuda::meanShiftSegmentation (InputArray src, OutputArray dst, int sp, int sr, int minsize, TermCriteria criteria=TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 5, 1), Stream &stream=Stream::Null())
	Performs a mean-shift segmentation of the source image and eliminates small segments.

Enumeration Type Documentation

ConnectedComponentsAlgorithmsTypes

enum cv::cuda::ConnectedComponentsAlgorithmsTypes

#include <opencv2/cudaimgproc.hpp>

Connected Components Algorithm.

Enumerator
CCL_DEFAULT	BKE [11] algorithm for 8-way connectivity.
CCL_BKE	BKE [11] algorithm for 8-way connectivity.

Function Documentation

bilateralFilter()

void cv::cuda::bilateralFilter	(	InputArray	src,
		OutputArray	dst,
		int	kernel_size,
		float	sigma_color,
		float	sigma_spatial,
		int	borderMode = BORDER_DEFAULT,
		Stream &	stream = Stream::Null()
	)

#include <opencv2/cudaimgproc.hpp>

Performs bilateral filtering of passed image.

Parameters

src	Source image. Supports only (channels != 2 && depth() != CV_8S && depth() != CV_32S && depth() != CV_64F).
dst	Destination imagwe.
kernel_size	Kernel window size.
sigma_color	Filter sigma in the color space.
sigma_spatial	Filter sigma in the coordinate space.
borderMode	Border type. See borderInterpolate for details. BORDER_REFLECT101 , BORDER_REPLICATE , BORDER_CONSTANT , BORDER_REFLECT and BORDER_WRAP are supported for now.
stream	Stream for the asynchronous version.

See also

bilateralFilter

blendLinear()

void cv::cuda::blendLinear	(	InputArray	img1,
		InputArray	img2,
		InputArray	weights1,
		InputArray	weights2,
		OutputArray	result,
		Stream &	stream = Stream::Null()
	)

#include <opencv2/cudaimgproc.hpp>

Performs linear blending of two images.

Parameters

img1	First image. Supports only CV_8U and CV_32F depth.
img2	Second image. Must have the same size and the same type as img1 .
weights1	Weights for first image. Must have tha same size as img1 . Supports only CV_32F type.
weights2	Weights for second image. Must have tha same size as img2 . Supports only CV_32F type.
result	Destination image.
stream	Stream for the asynchronous version.

connectedComponents() [1/2]

void cv::cuda::connectedComponents	(	InputArray	image,
		OutputArray	labels,
		int	connectivity,
		int	ltype,
		cv::cuda::ConnectedComponentsAlgorithmsTypes	ccltype
	)

#include <opencv2/cudaimgproc.hpp>

Computes the Connected Components Labeled image of a binary image.

The function takes as input a binary image and performs Connected Components Labeling. The output is an image where each Connected Component is assigned a unique label (integer value). ltype specifies the output label image type, an important consideration based on the total number of labels or alternatively the total number of pixels in the source image. ccltype specifies the connected components labeling algorithm to use, currently BKE [11] is supported, see the ConnectedComponentsAlgorithmsTypes for details. Note that labels in the output are not required to be sequential.

Parameters

image	The 8-bit single-channel image to be labeled.
labels	Destination labeled image.
connectivity	Connectivity to use for the labeling procedure. 8 for 8-way connectivity is supported.
ltype	Output image label type. Currently CV_32S is supported.
ccltype	Connected components algorithm type (see the ConnectedComponentsAlgorithmsTypes).

Note

A sample program demonstrating Connected Components Labeling in CUDA can be found at
opencv_contrib_source_code/modules/cudaimgproc/samples/connected_components.cpp

connectedComponents() [2/2]

void cv::cuda::connectedComponents	(	InputArray	image,
		OutputArray	labels,
		int	connectivity = 8,
		int	ltype = CV_32S
	)

#include <opencv2/cudaimgproc.hpp>

This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.

Parameters

image	The 8-bit single-channel image to be labeled.
labels	Destination labeled image.
connectivity	Connectivity to use for the labeling procedure. 8 for 8-way connectivity is supported.
ltype	Output image label type. Currently CV_32S is supported.

createCannyEdgeDetector()

Ptr< CannyEdgeDetector > cv::cuda::createCannyEdgeDetector	(	double	low_thresh,
		double	high_thresh,
		int	apperture_size = 3,
		bool	L2gradient = false
	)

#include <opencv2/cudaimgproc.hpp>

Creates implementation for cuda::CannyEdgeDetector .

Parameters

low_thresh	First threshold for the hysteresis procedure.
high_thresh	Second threshold for the hysteresis procedure.
apperture_size	Aperture size for the Sobel operator.
L2gradient	Flag indicating whether a more accurate \(L_2\) norm \(=\sqrt{(dI/dx)^2 + (dI/dy)^2}\) should be used to compute the image gradient magnitude ( L2gradient=true ), or a faster default \(L_1\) norm \(=|dI/dx|+|dI/dy|\) is enough ( L2gradient=false ).

createTemplateMatching()

Ptr< TemplateMatching > cv::cuda::createTemplateMatching	(	int	srcType,
		int	method,
		Size	user_block_size = Size()
	)

#include <opencv2/cudaimgproc.hpp>

Creates implementation for cuda::TemplateMatching .

Parameters

srcType	Input source type. CV_32F and CV_8U depth images (1..4 channels) are supported for now.
method	Specifies the way to compare the template with the image.
user_block_size	You can use field user_block_size to set specific block size. If you leave its default value Size(0,0) then automatic estimation of block size will be used (which is optimized for speed). By varying user_block_size you can reduce memory requirements at the cost of speed.

The following methods are supported for the CV_8U depth images for now:

• CV_TM_SQDIFF
• CV_TM_SQDIFF_NORMED
• CV_TM_CCORR
• CV_TM_CCORR_NORMED
• CV_TM_CCOEFF
• CV_TM_CCOEFF_NORMED

The following methods are supported for the CV_32F images for now:

• CV_TM_SQDIFF
• CV_TM_CCORR

See also

matchTemplate

meanShiftFiltering()

void cv::cuda::meanShiftFiltering	(	InputArray	src,
		OutputArray	dst,
		int	sp,
		int	sr,
		TermCriteria	criteria = TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 5, 1),
		Stream &	stream = Stream::Null()
	)

#include <opencv2/cudaimgproc.hpp>

Performs mean-shift filtering for each point of the source image.

Parameters

src	Source image. Only CV_8UC4 images are supported for now.
dst	Destination image containing the color of mapped points. It has the same size and type as src .
sp	Spatial window radius.
sr	Color window radius.
criteria	Termination criteria. See TermCriteria.
stream	Stream for the asynchronous version.

It maps each point of the source image into another point. As a result, you have a new color and new position of each point.

meanShiftProc()

void cv::cuda::meanShiftProc	(	InputArray	src,
		OutputArray	dstr,
		OutputArray	dstsp,
		int	sp,
		int	sr,
		TermCriteria	criteria = TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 5, 1),
		Stream &	stream = Stream::Null()
	)

#include <opencv2/cudaimgproc.hpp>

Performs a mean-shift procedure and stores information about processed points (their colors and positions) in two images.

Parameters

src	Source image. Only CV_8UC4 images are supported for now.
dstr	Destination image containing the color of mapped points. The size and type is the same as src .
dstsp	Destination image containing the position of mapped points. The size is the same as src size. The type is CV_16SC2 .
sp	Spatial window radius.
sr	Color window radius.
criteria	Termination criteria. See TermCriteria.
stream	Stream for the asynchronous version.

See also

cuda::meanShiftFiltering

meanShiftSegmentation()

void cv::cuda::meanShiftSegmentation	(	InputArray	src,
		OutputArray	dst,
		int	sp,
		int	sr,
		int	minsize,
		TermCriteria	criteria = TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 5, 1),
		Stream &	stream = Stream::Null()
	)

#include <opencv2/cudaimgproc.hpp>

Performs a mean-shift segmentation of the source image and eliminates small segments.

Parameters

src	Source image. Only CV_8UC4 images are supported for now.
dst	Segmented image with the same size and type as src (host or gpu memory).
sp	Spatial window radius.
sr	Color window radius.
minsize	Minimum segment size. Smaller segments are merged.
criteria	Termination criteria. See TermCriteria.
stream	Stream for the asynchronous version.