Image Geometry Transforms Functions

Routines manipulating an image geometry.

These functions can be found in the nppig library. Linking to only the sub-libraries that you use can significantly save link time, application load time, and CUDA runtime startup time when using dynamic libraries.

Geometric Transform API Specifics

This section covers some of the unique API features common to the geometric transform primitives.

Geometric Transforms and ROIs

Geometric transforms operate on source and destination ROIs. The way these ROIs affect the processing of pixels differs from other (non geometric) image-processing primitives: Only pixels in the intersection of the destination ROI and the transformed source ROI are being processed.

The typical processing proceedes as follows:

1. Transform the rectangular source ROI (given in source image coordinates) into the destination image space. This yields a quadrilateral.

2. Write only pixels in the intersection of the transformed source ROI and the destination ROI.

Pixel Interpolation

The majority of image geometry transform operation need to perform a resampling of the source image as source and destination pixels are not coincident.

NPP supports the following pixel inerpolation modes (in order from fastest to slowest and lowest to highest quality):

• nearest neighbor

• linear interpolation

• cubic convolution

• supersampling

• interpolation using Lanczos window function

Image Resize Square Pixel

ResizeSqrPixel

ResizeSqrPixel functions attempt to choose source pixels that would approximately represent the center of the destination pixels. It does so by using the following scaling formula to select source pixels for interpolation:

nAdjustedXFactor = 1.0 / nXFactor;
nAdjustedYFactor = 1.0 / nYFactor;
nAdjustedXShift = nXShift * nAdjustedXFactor + ((1.0 - nAdjustedXFactor) * 0.5);
nAdjustedYShift = nYShift * nAdjustedYFactor + ((1.0 - nAdjustedYFactor) * 0.5);
nSrcX = nAdjustedXFactor * nDstX - nAdjustedXShift;
nSrcY = nAdjustedYFactor * nDstY - nAdjustedYShift;

ResizeSqrPixel functions support the following interpolation modes:

NPPI_INTER_NN
NPPI_INTER_LINEAR
NPPI_INTER_CUBIC
NPPI_INTER_CUBIC2P_BSPLINE
NPPI_INTER_CUBIC2P_CATMULLROM
NPPI_INTER_CUBIC2P_B05C03
NPPI_INTER_SUPER
NPPI_INTER_LANCZOS

In the ResizeSqrPixel functions below source image clip checking is handled as follows:

If the source pixel fractional x and y coordinates are greater than or equal to oSizeROI.x and less than oSizeROI.x + oSizeROI.width and greater than or equal to oSizeROI.y and less than oSizeROI.y + oSizeROI.height then the source pixel is considered to be within the source image clip rectangle and the source image is sampled. Otherwise the source image is not sampled and a destination pixel is not written to the destination image.

Resize Error Codes

The resize primitives return the following error codes:

    - NPP_WRONG_INTERSECTION_ROI_ERROR indicates an error condition if
      srcROIRect has no intersection with the source image.
    - NPP_RESIZE_NO_OPERATION_ERROR if either destination ROI width or
      height is less than 1 pixel.
    - NPP_RESIZE_FACTOR_ERROR Indicates an error condition if either nXFactor or
      nYFactor is less than or equal to zero or in the case of NPPI_INTER_SUPER are not both downscaling.
    - NPP_INTERPOLATION_ERROR if eInterpolation has an illegal value.
    - NPP_SIZE_ERROR if source size width or height is less than 2 pixels.
    - NPP_RECTANGLE_ERROR if either destination ROI width or
      height is less than 1 pixel.

GetResizeRect

Returns NppiRect which represents the offset and size of the destination rectangle that would be generated by resizing the source NppiRect by the requested scale factors and shifts.

NppStatus nppiGetResizeRect(NppiRect oSrcROI, NppiRect *pDstRect, double nXFactor, double nYFactor, double nXShift, double nYShift, int eInterpolation)

Parameters

• oSrcROI – Region of interest in the source image.

• pDstRect – User supplied host memory pointer to an NppiRect structure that will be filled in by this function with the region of interest in the destination image.

• nXFactor – Factor by which x dimension is changed.

• nYFactor – Factor by which y dimension is changed.

• nXShift – Source pixel shift in x-direction.

• nYShift – Source pixel shift in y-direction.

• eInterpolation – The type of eInterpolation to perform resampling.

Returns

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

ResizeSqrPixel

Resizes images.

. Common parameters for nppiResizeSqrPixel packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F, NPP_64F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param nSrcStep

Source-Image Line Step.

param oSrcSize

Size in pixels of the source image.

param oSrcROI

Region of interest in the source image.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstROI

Region of interest in the destination image.

param nXFactor

Factor by which x dimension is changed.

param nYFactor

Factor by which y dimension is changed.

param nXShift

Source pixel shift in x-direction.

param nYShift

Source pixel shift in y-direction.

param eInterpolation

The type of eInterpolation to perform resampling.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

. Common parameters for nppiResizeSqrPixel planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F, NPP_64F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nSrcStep

Source-Image Line Step.

param oSrcSize

Size in pixels of the source image.

param oSrcROI

Region of interest in the source image.

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nDstStep

Destination-Image Line Step.

param oDstROI

Region of interest in the destination image.

param nXFactor

Factor by which x dimension is changed.

param nYFactor

Factor by which y dimension is changed.

param nXShift

Source pixel shift in x-direction.

param nYShift

Source pixel shift in y-direction.

param eInterpolation

The type of eInterpolation to perform resampling.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

NppStatus nppiResizeSqrPixel_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, void *pDst, int nDstStep, NppiRect oDstROI, double nXFactor, double nYFactor, double nXShift, double nYShift, int eInterpolation, NppStreamContext nppStreamCtx)

Image resize square pixel.

For common parameter descriptions, see . Common parameters for nppiResizeSqrPixel packed pixel functions include:.

NppStatus nppiResizeSqrPixel_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *const pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, void *pDst[4], int nDstStep, NppiRect oDstROI, double nXFactor, double nYFactor, double nXShift, double nYShift, int eInterpolation, NppStreamContext nppStreamCtx)

Planar image resize square pixel.

For common parameter descriptions, see /ref CommonResizeSqrPlanarPixelParameters.

NppStatus nppiResizeAdvancedGetBufferHostSize_8u_C1R(NppiSize oSrcROI, NppiSize oDstROI, int *hpBufferSize, int eInterpolationMode)

Buffer size for nppiResizeSqrPixel_Advanced.

Parameters

• oSrcROI – Region-of-Interest (ROI).

• oDstROI – Region-of-Interest (ROI).

• hpBufferSize – Required buffer size. Important: hpBufferSize is a host pointer. Scratch Buffer and Host Pointer.

• eInterpolationMode – The type of eInterpolation to perform resampling. Currently only supports NPPI_INTER_LANCZOS3_Advanced.

Returns

NPP_NULL_POINTER_ERROR if hpBufferSize is 0 (NULL), ROI Related Error Codes.

NppStatus nppiResizeSqrPixel_Advanced_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, int nSrcStep, NppiSize oSrcSize, NppiRect oSrcROI, void *pDst, int nDstStep, NppiRect oDstROI, double nXFactor, double nYFactor, Npp8u *pBuffer, int eInterpolationMode, NppStreamContext nppStreamCtx)

Image resize.

This primitive matches the behavior of GraphicsMagick++.

Supported data types include NPP_8U. Supported channel counts include NPP_CH_1.

Parameters

• eSrcDstType – image_data_type.

• eSrcDstChannels – image_channels.

• pSrc – Source-Image Pointer.

• nSrcStep – Source-Image Line Step.

• oSrcSize – Size in pixels of the source image.

• oSrcROI – Region of interest in the source image.

• pDst – Destination-Image Pointer.

• nDstStep – Destination-Image Line Step.

• oDstROI – Region of interest in the destination image.

• nXFactor – Factor by which x dimension is changed.

• nYFactor – Factor by which y dimension is changed.

• pBuffer – Device buffer that is used during calculations.

• eInterpolationMode – The type of eInterpolation to perform resampling. Currently only supports NPPI_INTER_LANCZOS3_Advanced.

• nppStreamCtx – Application Managed Stream Context.

Returns

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

Image Resize

Resize

Resize functions use scale factor automatically determined by the width and height ratios of input and output Region-of-Interest (ROI).

This simplified function replaces the previous version which was deprecated in an earlier release. In this function the resize scale factor is automatically determined by the width and height ratios of oSrcRectROI and oDstRectROI. If either of those parameters intersect their respective image sizes then pixels outside the image size width and height will not be processed.

Resize supports the following interpolation modes:

NPPI_INTER_NN
NPPI_INTER_LINEAR
NPPI_INTER_CUBIC
NPPI_INTER_SUPER
NPPI_INTER_LANCZOS

GetResizeTiledSourceOffset

Helper function that can be used when tiling a destination image with a source image using multiple Resize calls.

oSrcRectROI and oDstRectROI widths and heights should remain unmodified even if they will overlap source and destination image sizes. oDstRectROI offsets should be set to the destination offset of the new tile. Resize function processing will stop when source or destination image sizes are reached, any unavailable source image pixels beyond source image size will be border replicated. There is no particular association assumed between source and destination image locations. The values of oSrcRectROI.x and oSrcRectROI.y are ignored during this function call.

NppStatus nppiGetResizePlusTiledSourceOffset(NppiRect oSrcRectROI, NppiRect oDstRectROI, NppiPoint *pNewSrcRectOffset)

Parameters

• oSrcRectROI – Region of interest in the source image (may overlap source image size width and height).

• oDstRectROI – Region of interest in the destination image (may overlap destination image size width and height).

• pNewSrcRectOffset – Pointer to host memory NppiPoint object that will contain the new source image ROI offset to be used in the nppiResize call to generate that tile.

Returns

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

Resize

Resizes images.

. Common parameters for nppiResize packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_16F, NPP_32F, NPP_64F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer to origin of source image.

param nSrcStep

Source-Image Line Step.

param oSrcSize

Size in pixels of the entire source image.

param oSrcRectROI

Region of interest in the source image (may overlap source image size width and height).

param pDst

Destination-Image Pointer to origin of destination image.

param nDstStep

Destination-Image Line Step.

param oDstSize

Size in pixels of the entire destination image.

param oDstRectROI

Region of interest in the destination image (may overlap destination image size width and height).

param eInterpolation

The type of eInterpolation to perform resampling (16f versions do not support Lanczos interpolation).

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

. Common parameters for nppiResize planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F, NPP_64F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane origin pointers).

param nSrcStep

Source-Image Line Step.

param oSrcSize

Size in pixels of the entire source image.

param oSrcRectROI

Region of interest in the source image (may overlap source image size width and height).

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane origin pointers).

param nDstStep

Destination-Image Line Step.

param oDstSize

Size in pixels of the entire destination image.

param oDstRectROI

Region of interest in the destination image (may overlap destination image size width and height).

param eInterpolation

The type of eInterpolation to perform resampling.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

NppStatus nppiResize_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, int nSrcStep, NppiSize oSrcSize, NppiRect oSrcRectROI, void *pDst, int nDstStep, NppiSize oDstSize, NppiRect oDstRectROI, int eInterpolation, NppStreamContext nppStreamCtx)

Image resize.

For common parameter descriptions, see /ref CommonResizePackedPixelParameters.

NppStatus nppiResize_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *const pSrc[4], int nSrcStep, NppiSize oSrcSize, NppiRect oSrcRectROI, void *pDst[4], int nDstStep, NppiSize oDstSize, NppiRect oDstRectROI, int eInterpolation, NppStreamContext nppStreamCtx)

Planar image resize.

For common parameter descriptions, see /ref CommonResizePlanarPixelParameters.

Image Resize Batch

ResizeBatch

Above is the diagram of batch resize functions for variable ROIs. Figure shows the flexibility that the API can handle. The ROIs for source and destination images can be any rectangular width and height that reflects the needed resize factors, inside or beyond the image boundary.

ResizeBatch functions use scale factor automatically determined by the width and height ratios for each pair of input / output images in provided batches.

In this function as in nppiResize the resize scale factor is automatically determined by the width and height ratios of oSrcRectROI and oDstRectROI. If either of those parameters intersect their respective image sizes then pixels outside the image size width and height will not be processed. Details of the resize operation are described above in the Resize section. ResizeBatch generally takes the same parameter list as Resize except that there is a list of N instances of those parameters (N > 1) and that list is passed in device memory. A convenient data structure is provided that allows for easy initialization of the parameter lists. The only restriction on these functions is that there is one single source ROI rectangle and one single destination ROI rectangle which are applied respectively to each image in the batch. The primary purpose of this function is to provide improved performance for batches of smaller images as long as GPU resources are available. Therefore it is recommended that the function not be used for very large images as there may not be resources available for processing several large images simultaneously.

A single set of oSrcRectROI and oDstRectROI values are applied to each source image and destination image in the batch in the nppiResizeBatch version of the function while per image specific oSrcRectROI and oDstRectROI values can be used in the nppiResizeBatch_Advanced version of the function. Source and destination image sizes may vary but oSmallestSrcSize and oSmallestDstSize must be set to the smallest source and destination image sizes in the batch. The parameters in the

NppiResizeBatchCXR structure represent the corresponding per-image nppiResize parameters for each image in the nppiResizeBatch functions and the NppiImageDescriptor and NppiResizeBatchROI_Advanced structures represent the corresponding per-image nppiResize parameters for the nppiResizeBatch_Advanced functions. The NppiResizeBatchCXR or NppiImageDescriptor and NppiResizeBatchROI_Advanced arrays must be in device memory.

ResizeBatch supports the following interpolation modes:

NPPI_INTER_NN
NPPI_INTER_LINEAR
NPPI_INTER_CUBIC
NPPI_INTER_SUPER

Below is the diagram of batch resize functions for variable ROIs. Figure shows the flexibility that the API can handle. The ROIs for source and destination images can be any rectangular width and height that reflects the needed resize factors, inside or beyond the image boundary.

Common parameters for nppiResizeBatch functions include:

Supported data types include NPP_8U, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiResizeBatchAdvanced functions include:

Supported data types include NPP_8U, NPP_16F, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param oSmallestSrcSize

Size in pixels of the entire smallest source image width and height, may be from different images.

param oSrcRectROI

Region of interest in the source images (may overlap source image size width and height).

param oSmallestDstSize

Size in pixels of the entire smallest destination image width and height, may be from different images.

param oDstRectROI

Region of interest in the destination images (may overlap destination image size width and height).

param eInterpolation

The type of eInterpolation to perform resampling. Currently limited to NPPI_INTER_NN, NPPI_INTER_LINEAR, NPPI_INTER_CUBIC, or NPPI_INTER_SUPER.

param pBatchList

Device memory pointer to nBatchSize list of NppiResizeBatchCXR structures.

param pBatchSrc

Device pointer to NppiImageDescriptor list of source image descriptors. User needs to intialize this structure and copy it to device.

param pBatchDst

Device pointer to NppiImageDescriptor list of destination image descriptors. User needs to intialize this structure and copy it to device.

param pBatchROI

Device pointer to NppiResizeBatchROI_Advanced list of per-image variable ROIs. User needs to initialize this structure and copy it to device.

param nBatchSize

Number of NppiResizeBatchCXR structures in this call (must be > 1).

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param nMaxWidth

The maximum width of all destination ROIs

param nMaxHeight

The maximum height of all destination ROIs

param pBatchSrc

Device pointer to NppiImageDescriptor list of source image descriptors. User needs to intialize this structure and copy it to device.

param pBatchDst

Device pointer to NppiImageDescriptor list of destination image descriptors. User needs to intialize this structure and copy it to device.

param pBatchROI

Device pointer to NppiResizeBatchROI_Advanced list of per-image variable ROIs. User needs to initialize this structure and copy it to device.

param nBatchSize

Number of images in a batch.

param eInterpolation

The type of eInterpolation to perform resampling.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Resize Error Codes

Functions

NppStatus nppiResizeBatch_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, NppiSize oSmallestSrcSize, NppiRect oSrcRectROI, NppiSize oSmallestDstSize, NppiRect oDstRectROI, int eInterpolation, NppiResizeBatchCXR *pBatchList, unsigned int nBatchSize, NppStreamContext nppStreamCtx)

Image resize batch.

For common parameter descriptions, see /ref CommonResizeBatchParameters.

NppStatus nppiResizeBatch_Advanced_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, int nMaxWidth, int nMaxHeight, NppiImageDescriptor *pBatchSrc, NppiImageDescriptor *pBatchDst, NppiResizeBatchROI_Advanced *pBatchROI, unsigned int nBatchSize, int eInterpolation, NppStreamContext nppStreamCtx)

Image resize batch for variable ROI.

For common parameter descriptions, see /ref CommonResizeBatchAdvancedParameters.

Image Remap

Remap

Routines providing remap functionality.

Remap supports the following interpolation modes:

NPPI_INTER_NN NPPI_INTER_LINEAR NPPI_INTER_CUBIC NPPI_INTER_CUBIC2P_BSPLINE NPPI_INTER_CUBIC2P_CATMULLROM NPPI_INTER_CUBIC2P_B05C03 NPPI_INTER_LANCZOS

Remap chooses source pixels using pixel coordinates explicitely supplied in two 2D device memory image arrays pointed to by the pXMap and pYMap pointers. The pXMap array contains the X coordinated and the pYMap array contains the Y coordinate of the corresponding source image pixel to use as input. These coordinates are in floating point format so fraction pixel positions can be used. The coordinates of the source pixel to sample are determined as follows:

nSrcX = pxMap[nDstX, nDstY] nSrcY = pyMap[nDstX, nDstY]

In the Remap functions below source image clip checking is handled as follows:

If the source pixel fractional x and y coordinates are greater than or equal to oSizeROI.x and less than oSizeROI.x + oSizeROI.width and greater than or equal to oSizeROI.y and less than oSizeROI.y + oSizeROI.height then the source pixel is considered to be within the source image clip rectangle and the source image is sampled. Otherwise the source image is not sampled and a destination pixel is not written to the destination image.

Error Codes

The remap primitives return the following error codes:

    - NPP_WRONG_INTERSECTION_ROI_ERROR indicates an error condition if
      srcROIRect has no intersection with the source image.
    - NPP_INTERPOLATION_ERROR if eInterpolation has an illegal value.

Common parameters for nppiRemap packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiRemap planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param nSrcStep

Source-Image Line Step.

param oSrcSize

Size in pixels of the source image.

param oSrcROI

Region of interest in the source image.

param pXMap

Device memory pointer to 2D image array of X coordinate values to be used when sampling source image.

param nXMapStep

pXMap image array line step in bytes.

param pYMap

Device memory pointer to 2D image array of Y coordinate values to be used when sampling source image.

param nYMapStep

pYMap image array line step in bytes.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstSizeROI

Region of interest size in the destination image.

param eInterpolation

The type of eInterpolation to perform resampling.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array.

param nSrcStep

Source-Image Line Step.

param oSrcSize

Size in pixels of the source image.

param oSrcROI

Region of interest in the source image.

param pXMap

Device memory pointer to 2D image array of X coordinate values to be used when sampling source image.

param nXMapStep

pXMap image array line step in bytes.

param pYMap

Device memory pointer to 2D image array of Y coordinate values to be used when sampling source image.

param nYMapStep

pYMap image array line step in bytes.

param pDst

Destination-Planar-Image Pointer Array.

param nDstStep

Destination-Image Line Step.

param oDstSizeROI

Region of interest size in the destination image.

param eInterpolation

The type of eInterpolation to perform resampling.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Error Codes

Remap

Remaps images.

NppStatus nppiRemap_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp32f *pXMap, int nXMapStep, const Npp32f *pYMap, int nYMapStep, void *pDst, int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

Image remap.

For common parameter descriptions, see /ref CommonRemapPackedPixelParameters.

NppStatus nppiRemap_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *const pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp32f *pXMap, int nXMapStep, const Npp32f *pYMap, int nYMapStep, void *pDst[4], int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

Planar image remap.

For common parameter descriptions, see /ref CommonRemapPlanarPixelParameters.

NppStatus nppiRemap_64f_C1R_Ctx(const Npp64f *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp64f *pXMap, int nXMapStep, const Npp64f *pYMap, int nYMapStep, Npp64f *pDst, int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

1 channel 64-bit floating point image remap.

For common parameter descriptions, see /ref CommonRemapPackedPixelParameters.

NppStatus nppiRemap_64f_C3R_Ctx(const Npp64f *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp64f *pXMap, int nXMapStep, const Npp64f *pYMap, int nYMapStep, Npp64f *pDst, int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

3 channel 64-bit floating point image remap.

For common parameter descriptions, see /ref CommonRemapPackedPixelParameters.

NppStatus nppiRemap_64f_C4R_Ctx(const Npp64f *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp64f *pXMap, int nXMapStep, const Npp64f *pYMap, int nYMapStep, Npp64f *pDst, int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

4 channel 64-bit floating point image remap.

For common parameter descriptions, see /ref CommonRemapPackedPixelParameters.

NppStatus nppiRemap_64f_AC4R_Ctx(const Npp64f *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp64f *pXMap, int nXMapStep, const Npp64f *pYMap, int nYMapStep, Npp64f *pDst, int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

4 channel 64-bit floating point image remap not affecting alpha.

For common parameter descriptions, see /ref CommonRemapPackedPixelParameters.

NppStatus nppiRemap_64f_P3R_Ctx(const Npp64f *const pSrc[3], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp64f *pXMap, int nXMapStep, const Npp64f *pYMap, int nYMapStep, Npp64f *pDst[3], int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

3 channel 64-bit floating point planar image remap.

For common parameter descriptions, see /ref CommonRemapPlanarPixelParameters.

NppStatus nppiRemap_64f_P4R_Ctx(const Npp64f *const pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, const Npp64f *pXMap, int nXMapStep, const Npp64f *pYMap, int nYMapStep, Npp64f *pDst[4], int nDstStep, NppiSize oDstSizeROI, int eInterpolation, NppStreamContext nppStreamCtx)

4 channel 64-bit floating point planar image remap.

For common parameter descriptions, see /ref CommonRemapPlanarPixelParameters”.

Image Rotate

Rotate

Rotates an image around the origin (0,0) and then shifts it.

Rotate Error Codes

• NPP_INTERPOLATION_ERROR if eInterpolation has an illegal value.

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1.

• NPP_WRONG_INTERSECTION_ROI_ERROR indicates an error condition if srcROIRect has no intersection with the source image.

• NPP_WRONG_INTERSECTION_QUAD_WARNING indicates a warning that no operation is performed if the transformed source ROI does not intersect the destination ROI.

Image Rotate Utility Functions

Rotate Utility Functions

The set of rotate utility functions.

Functions

NppStatus nppiGetRotateQuad(NppiRect oSrcROI, double aQuad[4][2], double nAngle, double nShiftX, double nShiftY)

Compute shape of rotated image.

Parameters

• oSrcROI – Region-of-interest of the source image.

• aQuad – Array of 2D points. These points are the locations of the corners of the rotated ROI.

• nAngle – The rotation nAngle.

• nShiftX – Post-rotation shift in x-direction

• nShiftY – Post-rotation shift in y-direction

Returns

ROI Related Error Codes.

NppStatus nppiGetRotateBound(NppiRect oSrcROI, double aBoundingBox[2][2], double nAngle, double nShiftX, double nShiftY)

Compute bounding-box of rotated image.

Parameters

• oSrcROI – Region-of-interest of the source image.

• aBoundingBox – Two 2D points representing the bounding-box of the rotated image. All four points from nppiGetRotateQuad are contained inside the axis-aligned rectangle spanned by the the two points of this bounding box.

• nAngle – The rotation angle.

• nShiftX – Post-rotation shift in x-direction.

• nShiftY – Post-rotation shift in y-direction.

Returns

ROI Related Error Codes.

Image Mirror

Mirror

Mirrors images horizontally, vertically or diagonally.

Mirror Error Codes

• NPP_MIRROR_FLIP_ERROR if flip has an illegal value.

• NPP_SIZE_ERROR if in_place ROI width or height are not even numbers.

Common parameters for nppiMirror non-inplace and inplace functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrcDst

In-Place Image Pointer for inplace functions.

param nSrcDstStep

In-Place-Image Line Step for inplace functions.

param pSrc

Source-Image Pointer for non-inplace functions.

param nSrcStep

Source-Image Line Step for non-inplace functions.

param pDst

Destination-Image Pointer for non-inplace functions.

param nDstStep

Destination-Image Line Step for non-inplace functions.

param oROI

Region-of-Interest (ROI) (in_place ROI widths and heights must be even numbers).

param flip

Specifies the axis about which the image is to be mirrored.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Mirror Error Codes

Functions

NppStatus nppiMirror_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, int nSrcStep, void *pDst, int nDstStep, NppiSize oROI, NppiAxis flip, NppStreamContext nppStreamCtx)

Image mirror.

For common parameter descriptions, see /ref CommonMirrorParameters.

NppStatus nppiMirror_I_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, void *pSrcDst, int nSrcDstStep, NppiSize oROI, NppiAxis flip, NppStreamContext nppStreamCtx)

In place image mirror.

For common parameter descriptions, see /ref CommonMirrorParameters.

Image Affine Transforms

Affine Transforms

The set of affine transform functions available in the library.

Affine Transform Error Codes

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_WRONG_INTERSECTION_ROI_ERROR Indicates an error condition if oSrcROI has no intersection with the source image

• NPP_INTERPOLATION_ERROR Indicates an error condition if interpolation has an illegal value

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

• NPP_WRONG_INTERSECTION_QUAD_WARNING Indicates a warning that no operation is performed if the transformed source ROI has no intersection with the destination ROI

Image Affine Transform Utility Functions

Affine Transform Utility Functions

The set of affine transform utility functions.

Functions

NppStatus nppiGetAffineTransform(NppiRect oSrcROI, const double aQuad[4][2], double aCoeffs[2][3])

Computes affine transform coefficients based on source ROI and destination quadrilateral.

The function computes the coefficients of an affine transformation that maps the given source ROI (axis aligned rectangle with integer coordinates) to a quadrilateral in the destination image.

An affine transform in 2D is fully determined by the mapping of just three vertices. This function’s API allows for passing a complete quadrilateral effectively making the prolem overdetermined. What this means in practice is, that for certain quadrilaterals it is not possible to find an affine transform that would map all four corners of the source ROI to the four vertices of that quadrilateral.

The function circumvents this problem by only looking at the first three vertices of the destination image quadrilateral to determine the affine transformation’s coefficients. If the destination quadrilateral is indeed one that cannot be mapped using an affine transformation the functions informs the user of this situation by returning a NPP_AFFINE_QUAD_INCORRECT_WARNING.

Parameters

• oSrcROI – The source ROI. This rectangle needs to be at least one pixel wide and high. If either width or hight are less than one an NPP_RECTANGLE_ERROR is returned.

• aQuad – The destination quadrilateral.

• aCoeffs – The resulting affine transform coefficients.

Returns

Error codes:

• NPP_SIZE_ERROR Indicates an error condition if any image dimension has zero or negative value

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

• NPP_AFFINE_QUAD_INCORRECT_WARNING Indicates a warning when quad does not conform to the transform properties. Fourth vertex is ignored, internally computed coordinates are used instead

NppStatus nppiGetAffineQuad(NppiRect oSrcROI, double aQuad[4][2], const double aCoeffs[2][3])

Compute shape of transformed image.

This method computes the quadrilateral in the destination image that the source ROI is transformed into by the affine transformation expressed by the coefficients array (aCoeffs).

Parameters

• oSrcROI – The source ROI.

• aQuad – The resulting destination quadrangle.

• aCoeffs – The afine transform coefficients.

Returns

Error codes:

• NPP_SIZE_ERROR Indicates an error condition if any image dimension has zero or negative value

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

NppStatus nppiGetAffineBound(NppiRect oSrcROI, double aBound[2][2], const double aCoeffs[2][3])

Compute bounding-box of transformed image.

The method effectively computes the bounding box (axis aligned rectangle) of the transformed source ROI (see nppiGetAffineQuad()).

Parameters

• oSrcROI – The source ROI.

• aBound – The resulting bounding box.

• aCoeffs – The afine transform coefficients.

Returns

Error codes:

• NPP_SIZE_ERROR Indicates an error condition if any image dimension has zero or negative value

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

Image Affine Transform

Affine Transform

Transforms (warps) an image based on an affine transform.

The affine transform is given as a \(2\times 3\) matrix C. A pixel location \((x, y)\) in the source image is mapped to the location \((x', y')\) in the destination image. The destination image coorodinates are computed as follows:

\[ x' = c_{00} * x + c_{01} * y + c_{02} \qquad y' = c_{10} * x + c_{11} * y + c_{12} \qquad C = \left[ \matrix{c_{00} & c_{01} & c_{02} \cr c_{10} & c_{11} & c_{12} } \right] \]

Affine transforms can be understood as a linear transformation (traditional matrix multiplication) and a shift operation. The \(2\times 2\) matrix

\[ L = \left[ \matrix{c_{00} & c_{01} \cr c_{10} & c_{11} } \right] \]

represents the linear transform portion of the affine transformation. The vector

\[ v = \left( \matrix{c_{02} \cr c_{12} } \right) \]

represents the post-transform shift, i.e. after the pixel location is transformed by \(L\) it is translated by \(v\).

Common parameters for nppiWarpAffine packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_16F, NPP_32F, NPP_64F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiWarpAffine planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F, NPP_64F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Affine transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Affine Transform Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Planar-Image Pointer Array. (host memory array containing device memory image plane pointers)

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Affine transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Affine Transform Error Codes

Functions

NppStatus nppiWarpAffine_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, void *pDst, int nDstStep, NppiRect oDstRectROI, const double aCoeffs[2][3], int eInterpolation, NppStreamContext nppStreamCtx)

Affine warp.

For common parameter descriptions, see /ref CommonWarpAffinePackedPixelParameters.

NppStatus nppiWarpAffine_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, void *pDst[4], int nDstStep, NppiRect oDstRectROI, const double aCoeffs[2][3], int eInterpolation, NppStreamContext nppStreamCtx)

Planar affine warp.

For common parameter descriptions, see /ref CommonWarpAffinePlanarPixelParameters.

Image Affine Transform Batch

Affine Transform Batch

Details of the warp affine operation are described above in the WarpAffine section. WarpAffineBatch generally takes the same parameter list as WarpAffine except that there is a list of N instances of those parameters (N > 1) and that list is passed in device memory. A convenient data structure is provided that allows for easy initialization of the parameter lists. The aTransformedCoeffs array is for internal use only and should not be directly initialized by the application. The only restriction on these functions is that there is one single source ROI rectangle and one single destination ROI rectangle which are applied respectively to each image in the batch. The primary purpose of this function is to provide improved performance for batches of smaller images as long as GPU resources are available. Therefore it is recommended that the function not be used for very large images as there may not be resources available for processing several large images simultaneously.

A single set of oSrcRectROI and oDstRectROI values are applied to each source image and destination image in the batch. Source and destination image sizes may vary but oSmallestSrcSize must be set to the smallest source and image size in the batch. The parameters in the

NppiWarpAffineBatchCXR structure represent the corresponding per-image nppiWarpAffine parameters for each image in the batch. The NppiWarpAffineBatchCXR array must be in device memory. The nppiWarpAffineBatchInit function MUST be called AFTER the application has initialized the array of NppiWarpAffineBatchCXR structures and BEFORE calling any of the nppiWarpAffineBatch functions to so that the aTransformedCoeffs array can be internally pre-initialized for each image in the batch. The batch size passed to nppiWarpAffineBatchInit must match the batch size passed to the corresponding warp affine batch function.

WarpAffineBatch supports the following interpolation modes:

NPPI_INTER_NN
NPPI_INTER_LINEAR
NPPI_INTER_CUBIC

Codes

The warp affine primitives return the following error codes:

    - NPP_RECTANGLE_ERROR if either destination ROI width or
      height is less than 1 pixel.
    - NPP_INTERPOLATION_ERROR if eInterpolation has an illegal value.
    - NPP_SIZE_ERROR if source size width or height is less than 2 pixels.

Common parameters for nppiWarpAffineBatch functions include:

Supported data types include NPP_8U, NPP_16F, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param oSmallestSrcSize

Size in pixels of the entire smallest source image width and height, may be from different images.

param oSrcRectROI

Region of interest in the source images.

param oDstRectROI

Region of interest in the destination images.

param eInterpolation

The type of eInterpolation to perform resampling. Currently limited to NPPI_INTER_NN, NPPI_INTER_LINEAR, or NPPI_INTER_CUBIC.

param pBatchList

Device memory pointer to nBatchSize list of NppiWarpAffineBatchCXR structures.

param nBatchSize

Number of NppiWarpAffineBatchCXR structures in this call (must be > 1).

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes

Functions

NppStatus nppiWarpAffineBatchInit_Ctx(NppiWarpAffineBatchCXR *pBatchList, unsigned int nBatchSize, NppStreamContext nppStreamCtx)

Initializes the aTransformdedCoeffs array in pBatchList for each image in the list.

MUST be called before calling the corresponding warp affine batch function whenever any of the transformation matrices in the list have changed.

Parameters

• pBatchList – Device memory pointer to nBatchSize list of NppiWarpAffineBatchCXR structures.

• nBatchSize – Number of NppiWarpAffineBatchCXR structures in this call (must be > 1).

• nppStreamCtx – Application Managed Stream Context.

NppStatus nppiWarpAffineBatch_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, NppiSize oSmallestSrcSize, NppiRect oSrcRectROI, NppiRect oDstRectROI, int eInterpolation, NppiWarpAffineBatchCXR *pBatchList, unsigned int nBatchSize, NppStreamContext nppStreamCtx)

Image warp affine batch.

For common parameter descriptions, see /ref CommonWarpAffineBatchParameters.

Image Backwards Affine Transform

Backwards Affine Transform

Transforms (warps) an image based on an affine transform.

The affine transform is given as a \(2\times 3\) matrix C. A pixel location \((x, y)\) in the source image is mapped to the location \((x', y')\) in the destination image. The destination image coorodinates fullfil the following properties:

\[ x = c_{00} * x' + c_{01} * y' + c_{02} \qquad y = c_{10} * x' + c_{11} * y' + c_{12} \qquad C = \left[ \matrix{c_{00} & c_{01} & c_{02} \cr c_{10} & c_{11} & c_{12} } \right] \]

In other words, given matrix \(C\) the source image’s shape is transfored to the destination image using the inverse matrix \(C^{-1}\):

\[ M = C^{-1} = \left[ \matrix{m_{00} & m_{01} & m_{02} \cr m_{10} & m_{11} & m_{12} } \right] x' = m_{00} * x + m_{01} * y + m_{02} \qquad y' = m_{10} * x + m_{11} * y + m_{12} \qquad \]

. Common parameters for nppiWarpAffineBack packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_16F, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiWarpAffineBack planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Affine transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Affine Transform Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Affine transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Affine Transform Error Codes

Functions

NppStatus nppiWarpAffineBack_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcrectROI, void *pDst, int nDstStep, NppiRect oDstRectROI, const double aCoeffs[2][3], int eInterpolation, NppStreamContext nppStreamCtx)

Image backwards affine warp.

For common parameter descriptions, see /ref CommonWarpAffineBackPackedPixelParameters.

NppStatus nppiWarpAffineBack_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, void *pDst[4], int nDstStep, NppiRect oDstROI, const double aCoeffs[2][3], int eInterpolation, NppStreamContext nppStreamCtx)

Planar image backwards affine warp.

For common parameter descriptions, see /ref CommonWarpAffineBackPlanarPixelParameters.

Image Quad Based Affine Transform

Quad-Based Affine Transform

Transforms (warps) an image based on an affine transform.

The affine transform is computed such that it maps a quadrilateral in source image space to a quadrilateral in destination image space.

An affine transform is fully determined by the mapping of 3 discrete points. The following primitives compute an affine transformation matrix that maps the first three corners of the source quad are mapped to the first three vertices of the destination image quad. If the fourth vertices do not match the transform, an NPP_AFFINE_QUAD_INCORRECT_WARNING is returned by the primitive.

Common parameters for nppiWarpAffineQuad packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiWarpAffineQuad planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param aSrcQuad

Source quad.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aDstQuad

Destination quad.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Affine Transform Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI

param aSrcQuad

Source quad.

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI

param aDstQuad

Destination quad.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Affine Transform Error Codes

Functions

NppStatus nppiWarpAffineQuad_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, const double aSrcQuad[4][2], void *pDst, int nDstStep, NppiRect oDstRectROI, const double aDstQuad[4][2], int eInterpolation, NppStreamContext nppStreamCtx)

Image quad-based affine warp.

For common parameter descriptions, see /ref CommonWarpAffineQuadPackedPixelParameters.

NppStatus nppiWarpAffineQuad_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, const double aSrcQuad[4][2], void *pDst[4], int nDstStep, NppiRect oDstRectROI, const double aDstQuad[4][2], int eInterpolation, NppStreamContext nppStreamCtx)

Planar image quad-based affine warp.

For common parameter descriptions, see /ref CommonWarpAffineQuadPlanarPixelParameters.

Image Perspective Transforms

Perspective Transform

The set of perspective transform functions available in the library.

Perspective Transform Error Codes

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_WRONG_INTERSECTION_ROI_ERROR Indicates an error condition if oSrcROI has no intersection with the source image

• NPP_INTERPOLATION_ERROR Indicates an error condition if interpolation has an illegal value

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

• NPP_WRONG_INTERSECTION_QUAD_WARNING Indicates a warning that no operation is performed if the transformed source ROI has no intersection with the destination ROI

Image Perspective Transform Utility Functions

Perspective Transform Utility Functions

The set of perspective transform utility functions.

Functions

NppStatus nppiGetPerspectiveTransform(NppiRect oSrcROI, const double quad[4][2], double aCoeffs[3][3])

Calculates perspective transform coefficients given source rectangular ROI and its destination quadrangle projection.

Parameters

• oSrcROI – Source ROI

• quad – Destination quadrangle

• aCoeffs – Perspective transform coefficients

Returns

Error codes:

• NPP_SIZE_ERROR Indicates an error condition if any image dimension has zero or negative value

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

NppStatus nppiGetPerspectiveQuad(NppiRect oSrcROI, double quad[4][2], const double aCoeffs[3][3])

Calculates perspective transform projection of given source rectangular ROI.

Parameters

• oSrcROI – Source ROI

• quad – Destination quadrangle

• aCoeffs – Perspective transform coefficients

Returns

Error codes:

• NPP_SIZE_ERROR Indicates an error condition if any image dimension has zero or negative value

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

NppStatus nppiGetPerspectiveBound(NppiRect oSrcROI, double bound[2][2], const double aCoeffs[3][3])

Calculates bounding box of the perspective transform projection of the given source rectangular ROI.

Parameters

• oSrcROI – Source ROI

• bound – Bounding box of the transformed source ROI

• aCoeffs – Perspective transform coefficients

Returns

Error codes:

• NPP_SIZE_ERROR Indicates an error condition if any image dimension has zero or negative value

• NPP_RECTANGLE_ERROR Indicates an error condition if width or height of the intersection of the oSrcROI and source image is less than or equal to 1

• NPP_COEFFICIENT_ERROR Indicates an error condition if coefficient values are invalid

Image Perspective Transform

Perspective Transform

Transforms (warps) an image based on a perspective transform.

The perspective transform is given as a \(3\times 3\) matrix C. A pixel location \((x, y)\) in the source image is mapped to the location \((x', y')\) in the destination image. The destination image coorodinates are computed as follows:

\[ x' = \frac{c_{00} * x + c_{01} * y + c_{02}}{c_{20} * x + c_{21} * y + c_{22}} \qquad y' = \frac{c_{10} * x + c_{11} * y + c_{12}}{c_{20} * x + c_{21} * y + c_{22}} \]

\[ C = \left[ \matrix{c_{00} & c_{01} & c_{02} \cr c_{10} & c_{11} & c_{12} \cr c_{20} & c_{21} & c_{22} } \right] \]

Common parameters for nppiWarpPerspective packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_16F, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiWarpPerspective planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Perspective transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Perspective Transform Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Perspective transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Perspective Transform Error Codes

Functions

NppStatus nppiWarpPerspective_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, void *pDst, int nDstStep, NppiRect oDstRectROI, const double aCoeffs[3][3], int eInterpolation, NppStreamContext nppStreamCtx)

Image perspective warp.

For common parameter descriptions, see /ref CommonWarpPerspectivePackedPixelParameters.

NppStatus nppiWarpPerspective_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, void *pDst[4], int nDstStep, NppiRect oDstRectROI, const double aCoeffs[2][3], int eInterpolation, NppStreamContext nppStreamCtx)

Planar image perspective warp.

For common parameter descriptions, see /ref CommonWarpPerspectivePlanarPixelParameters.

Image Perspective Transform Batch

Perspective Transform Batch

Details of the warp perspective operation are described above in the WarpPerspective section. WarpPerspectiveBatch generally takes the same parameter list as WarpPerspective except that there is a list of N instances of those parameters (N > 1) and that list is passed in device memory. A convenient data structure is provided that allows for easy initialization of the parameter lists. The aTransformedCoeffs array is for internal use only and should not be directly initialized by the application. The only restriction on these functions is that there is one single source ROI rectangle and one single destination ROI rectangle which are applied respectively to each image in the batch. The primary purpose of this function is to provide improved performance for batches of smaller images as long as GPU resources are available. Therefore it is recommended that the function not be used for very large images as there may not be resources available for processing several large images simultaneously.

A single set of oSrcRectROI and oDstRectROI values are applied to each source image and destination image in the batch. Source and destination image sizes may vary but oSmallestSrcSize must be set to the smallest source and image size in the batch. The parameters in the

NppiWarpPerspectiveBatchCXR structure represent the corresponding per-image nppiWarpPerspective parameters for each image in the batch. The NppiWarpPerspectiveBatchCXR array must be in device memory. The nppiWarpPerspectiveBatchInit function MUST be called AFTER the application has initialized the array of NppiWarpPerspectiveBatchCXR structures and BEFORE calling any of the nppiWarpPerspectiveBatch functions to so that the aTransformedCoeffs array can be internally pre-initialized for each image in the batch. The batch size passed to nppiWarpPerspectiveBatchInit must match the batch size passed to the corresponding warp perspective batch function.

WarpPerspectiveBatch supports the following interpolation modes:

NPPI_INTER_NN
NPPI_INTER_LINEAR
NPPI_INTER_CUBIC

Common parameters for nppiWarpPerspectiveBatch functions include:

Supported data types include NPP_8U, NPP_16F, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param oSmallestSrcSize

Size in pixels of the entire smallest source image width and height, may be from different images.

param oSrcRectROI

Region of interest in the source images.

param oDstRectROI

Region of interest in the destination images.

param eInterpolation

The type of eInterpolation to perform resampling. Currently limited to NPPI_INTER_NN, NPPI_INTER_LINEAR, or NPPI_INTER_CUBIC.

param pBatchList

Device memory pointer to nBatchSize list of NppiWarpPerspectiveBatchCXR structures.

param nBatchSize

Number of NppiWarpPerspectiveBatchCXR structures in this call (must be > 1).

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes

Functions

NppStatus nppiWarpPerspectiveBatchInit_Ctx(NppiWarpPerspectiveBatchCXR *pBatchList, unsigned int nBatchSize, NppStreamContext nppStreamCtx)

Initializes the aTransformdedCoeffs array in pBatchList for each image in the list.

MUST be called before calling the corresponding warp perspective batch function whenever any of the transformation matrices in the list have changed.

Parameters

• pBatchList – Device memory pointer to nBatchSize list of NppiWarpPerspectiveBatchCXR structures.

• nBatchSize – Number of NppiWarpPerspectiveBatchCXR structures in this call (must be > 1).

• nppStreamCtx – Application Managed Stream Context.

NppStatus nppiWarpPerspectiveBatch_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, NppiSize oSmallestSrcSize, NppiRect oSrcRectROI, NppiRect oDstRectROI, int eInterpolation, NppiWarpPerspectiveBatchCXR *pBatchList, unsigned int nBatchSize, NppStreamContext nppStreamCtx)

Image warp perspective batch.

For common parameter descriptions, see /ref CommonWarpPerspectiveBatchParameters.

Image Backwards Perspective Transform

Backwards Perspective Transform

Transforms (warps) an image based on a perspective transform.

The perspective transform is given as a \(3\times 3\) matrix C. A pixel location \((x, y)\) in the source image is mapped to the location \((x', y')\) in the destination image. The destination image coorodinates fullfil the following properties:

\[ x = \frac{c_{00} * x' + c_{01} * y' + c_{02}}{c_{20} * x' + c_{21} * y' + c_{22}} \qquad y = \frac{c_{10} * x' + c_{11} * y' + c_{12}}{c_{20} * x' + c_{21} * y' + c_{22}} \]

\[ C = \left[ \matrix{c_{00} & c_{01} & c_{02} \cr c_{10} & c_{11} & c_{12} \cr c_{20} & c_{21} & c_{22} } \right] \]

In other words, given matrix \(C\) the source image’s shape is transfored to the destination image using the inverse matrix \(C^{-1}\):

\[ M = C^{-1} = \left[ \matrix{m_{00} & m_{01} & m_{02} \cr m_{10} & m_{11} & m_{12} \cr m_{20} & m_{21} & m_{22} } \right] x' = \frac{c_{00} * x + c_{01} * y + c_{02}}{c_{20} * x + c_{21} * y + c_{22}} \qquad y' = \frac{c_{10} * x + c_{11} * y + c_{12}}{c_{20} * x + c_{21} * y + c_{22}} \]

Common parameters for nppiWarpPerspectiveBack packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiWarpPerspectiveBack planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Perspective transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Perspective Transform Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aCoeffs

Perspective transform coefficients.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Perspective Transform Error Codes

Functions

NppStatus nppiWarpPerspectiveBack_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcROI, void *pDst, int nDstStep, NppiRect oDstROI, const double aCoeffs[3][3], int eInterpolation, NppStreamContext nppStreamCtx)

Image backwards perspective warp.

For common parameter descriptions, see /ref CommonWarpPerspectiveBackPackedPixelParameters.

NppStatus nppiWarpPerspectiveBack_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, void *pDst[4], int nDstStep, NppiRect oDstRectROI, const double aCoeffs[3][3], int eInterpolation, NppStreamContext nppStreamCtx)

Planar image backwards perspective warp.

For common parameter descriptions, see /ref CommonWarpPerspectiveBackPlanarPixelParameters.

Image Quad Based Perspective Transform

Quad-Based Perspective Transform

Transforms (warps) an image based on an perspective transform.

The perspective transform is computed such that it maps a quadrilateral in source image space to a quadrilateral in destination image space.

Common parameters for nppiWarpPerspectiveQuad packed pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_16S, NPP_32F. Supported channel counts include NPP_CH_1, NPP_CH_3, NPP_CH_4, NPP_CH_A4.

Common parameters for nppiWarpPerspectiveQuad planar pixel functions include:

Supported data types include NPP_8U, NPP_16U, NPP_32S, NPP_32F. Supported channel counts include NPP_CH_P3, NPP_CH_P4.

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Image Pointer.

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param aSrcQuad

Source quad.

param pDst

Destination-Image Pointer.

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aDstQuad

Destination quad.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Perspective Transform Error Codes

param eSrcDstType

image_data_type.

param eSrcDstChannels

image_channels.

param pSrc

Source-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param oSrcSize

Size of source image in pixels.

param nSrcStep

Source-Image Line Step.

param oSrcRectROI

Source ROI.

param aSrcQuad

Source quad.

param pDst

Destination-Planar-Image Pointer Array (host memory array containing device memory image plane pointers).

param nDstStep

Destination-Image Line Step.

param oDstRectROI

Destination ROI.

param aDstQuad

Destination quad.

param eInterpolation

Interpolation mode: can be NPPI_INTER_NN, NPPI_INTER_LINEAR or NPPI_INTER_CUBIC.

param nppStreamCtx

Application Managed Stream Context.

return

Image Data Related Error Codes, ROI Related Error Codes, Perspective Transform Error Codes

Functions

NppStatus nppiWarpPerspectiveQuad_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc, NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, const double aSrcQuad[4][2], void *pDst, int nDstStep, NppiRect oDstRectROI, const double aDstQuad[4][2], int eInterpolation, NppStreamContext nppStreamCtx)

Image Quad-based perspective warp.

For common parameter descriptions, see /ref CommonWarpPerspectiveQuadPackedPixelParameters.

NppStatus nppiWarpPerspectiveQuad_Planar_Ctx(NppDataType eSrcDstType, NppiChannels eSrcDstChannels, const void *pSrc[4], NppiSize oSrcSize, int nSrcStep, NppiRect oSrcRectROI, const double aSrcQuad[4][2], void *pDst[4], int nDstStep, NppiRect oDstRectROI, const double aDstQuad[4][2], int eInterpolation, NppStreamContext nppStreamCtx)

In place image Quad-based perspective warp.

For common parameter descriptions, see /ref CommonWarpPerspectiveQuadPlanarPixelParameters.