Vision Primitives API

Detailed Description

This section includes the NVIDIA vision extensions supported in VisionWorks.

Target device possibilities for primitives

Primitive	GPU	CPU (optimized for ARM only)	CPU (optimized)	CPU
VX_KERNEL_COLOR_CONVERT	X	X (RGB[X]->Gray, RGB[X]<->RGB[X])		X
VX_KERNEL_CHANNEL_EXTRACT	X	X		X
VX_KERNEL_CHANNEL_COMBINE	X	X		X
VX_KERNEL_SOBEL_3x3	X	X		X
VX_KERNEL_MAGNITUDE	X	X		X
VX_KERNEL_PHASE	X	X		X
VX_KERNEL_SCALE_IMAGE	X	X (except VX_BORDER_MODE_CONSTANT)		X
VX_KERNEL_TABLE_LOOKUP	X		X
VX_KERNEL_HISTOGRAM	X		X
VX_KERNEL_EQUALIZE_HISTOGRAM	X		X
VX_KERNEL_ABSDIFF	X	X		X
VX_KERNEL_MEAN_STDDEV	X	X		X
VX_KERNEL_THRESHOLD	X	X		X
VX_KERNEL_INTEGRAL_IMAGE	X	X		X
VX_KERNEL_DILATE_3x3	X	X		X
VX_KERNEL_ERODE_3x3	X	X		X
VX_KERNEL_MEDIAN_3x3	X	X (except VX_BORDER_MODE_CONSTANT)		X
VX_KERNEL_BOX_3x3	X	X		X
VX_KERNEL_GAUSSIAN_3x3	X	X		X
VX_KERNEL_CUSTOM_CONVOLUTION	X	X		X
VX_KERNEL_GAUSSIAN_PYRAMID	X	X		X
VX_KERNEL_ACCUMULATE	X	X		X
VX_KERNEL_ACCUMULATE_WEIGHTED	X	X		X
VX_KERNEL_ACCUMULATE_SQUARE	X	X		X
VX_KERNEL_MINMAXLOC	X	X		X
VX_KERNEL_CONVERTDEPTH	X	X		X
VX_KERNEL_CANNY_EDGE_DETECTOR	X			X
VX_KERNEL_AND	X	X		X
VX_KERNEL_OR	X	X		X
VX_KERNEL_XOR	X	X		X
VX_KERNEL_NOT	X	X		X
VX_KERNEL_MULTIPLY	X	X (VX_ROUND_POLICY_TO_ZERO only)		X
VX_KERNEL_ADD	X	X		X
VX_KERNEL_SUBTRACT	X	X		X
VX_KERNEL_WARP_AFFINE	X	X (VX_DF_IMAGE_U8 only)		X
VX_KERNEL_WARP_PERSPECTIVE	X	X (VX_DF_IMAGE_U8 only)		X
VX_KERNEL_HARRIS_CORNERS	X			X
VX_KERNEL_FAST_CORNERS	X			X
VX_KERNEL_OPTICAL_FLOW_PYR_LK	X	X		X
VX_KERNEL_REMAP	X	X		X
VX_KERNEL_HALFSCALE_GAUSSIAN	X			X
NVX_KERNEL_HARRIS_TRACK	X
NVX_KERNEL_FAST_TRACK	X		X
NVX_KERNEL_FLIP_IMAGE	X	X		X
NVX_KERNEL_COPY_IMAGE	X		X
NVX_KERNEL_SEMI_GLOBAL_MATCHING	X
NVX_KERNEL_HOUGH_LINES	X
NVX_KERNEL_HOUGH_SEGMENTS	X
NVX_KERNEL_HOUGH_CIRCLES	X		X
NVX_KERNEL_SCHARR_3x3	X			X
NVX_KERNEL_LAPLACIAN_3x3	X	X		X
NVX_KERNEL_MEDIAN_FLOW			X
NVX_KERNEL_STEREO_BLOCK_MATCHING	X			X
NVX_KERNEL_FIND_HOMOGRAPHY			X
NVX_KERNEL_CREATE_MOTION_FIELD	X			X
NVX_KERNEL_REFINE_MOTION_FIELD	X			X
NVX_KERNEL_PARTITION_MOTION_FIELD	X			X
NVX_KERNEL_MULTIPLY_BY_SCALAR	X			X

Modules
	Copy Image
	Copies data from one image to another.
	FAST Track
	The FAST Track Primitive.
	Find Homography
	The Find Homography Algorithm.
	Harris Track
	The Harris Track Primitive.
	Flip Image
	The Flip Image Primitive.
	Color Convert Extensions
	Adds various extensions to the Color Convert vision function.
	Scale Image Extensions
	Adds various extensions to the Scale Image vision function.
	Warp Extensions
	Adds various extensions to the WarpAffine and WarpPerspective vision functions.
	Harris Corners Extensions
	Adds various extensions to the Harris Corners vision function.
	Fast Corners Extensions
	Adds various extensions to the Fast Corners vision function.
	Optical Flow PyrLK Extensions
	Adds various extensions to the Optical Flow PyrLK vision function.
	Pixel-wise Multiplication Extensions
	Adds various extensions to the Pixel-wise Multiplication vision function.
	Gaussian Pyramid Extensions
	Adds various extensions to the Gaussian Pyramid vision function.
	Semi-Global Matching
	The Semi-Global Matching Primitive is used for disparity map reconstruction.
	Hough Circles
	The Hough Circles primitive finds circles in a binary image using a modification of the Hough transform.
	Hough Lines
	The Hough Lines primitive finds lines in a binary image using Hough transform.
	Scharr3x3
	The Scharr 3x3 Image Filter Kernel.
	Laplacian3x3
	The Laplacian3x3 operator.
	Median Flow
	The Median Flow Algorithm.
	Stereo Block Matching
	The Stereo Block Matching Primitive.
	Multiply by Scalar
	Multiplies an input image by scalar and optionally converts it to another data type.
	IME : Create Motion Field
	Creates initial motion field from a current image into reference image.
	IME : Refine Motion Field
	Iteratively refines motion field by applying the motion vectors to a center block in a 3x3 block neighborhood.
	IME : Partition Motion Field
	Partitions motion vector field for blocks onto motion vector field into quarter sized blocks (half size in each dimension).