Package net.preibisch.mvrecon.process.cuda

Class CUDASeparableConvolutionFunctions

java.lang.Object

net.preibisch.mvrecon.process.cuda.CUDASeparableConvolutionFunctions

public class CUDASeparableConvolutionFunctions
extends Object

Executes gaussian convolution using the native CUDA implementation (https://github.com/StephanPreibisch/SeparableConvolutionCUDALib)

Author:

Stephan Preibisch (stephan.preibisch@gmx.de)

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	CUDASeparableConvolutionFunctions.OutOfBounds

Constructor Summary

Constructors

Constructor	Description
CUDASeparableConvolutionFunctions(CUDASeparableConvolution cuda, int cudaDeviceId)

Method Summary

Modifier and Type	Method	Description
boolean	gauss(float[] img, int[] dim, double sigma)	Compute Gaussian convolution with and extend-border-pixels outofbounds strategy
boolean	gauss(float[] img, int[] dim, double[] sigma)	Compute Gaussian convolution with and extend-border-pixels outofbounds strategy
boolean	gauss(float[] img, int[] dim, double[] sigma, CUDASeparableConvolutionFunctions.OutOfBounds oobs, float oobsValue)	Compute Gaussian convolution
static final boolean	gauss(float[] img, int[] dim, double[] sigma, CUDASeparableConvolutionFunctions.OutOfBounds oobs, float oobsValue, CUDASeparableConvolution cuda, int cudaDeviceId)	Compute Gaussian convolution
boolean	gauss(float[] img, int[] dim, double sigma, CUDASeparableConvolutionFunctions.OutOfBounds oobs, float oobsValue)	Compute Gaussian convolution
static final boolean	gauss(float[] img, int[] dim, double sigma, CUDASeparableConvolutionFunctions.OutOfBounds oobs, float oobsValue, CUDASeparableConvolution cuda, int cudaDeviceId)	Compute Gaussian convolution
boolean	gauss(float[] img, int[] dim, float[] sigma)
static float[][]	getCUDAKernels(double[] sigma, int[] supportedKernelSizes)
static float[]	getFloatKernelPadded(double[] kernel, int size)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

CUDASeparableConvolutionFunctions

public CUDASeparableConvolutionFunctions(CUDASeparableConvolution cuda,
 int cudaDeviceId)

Method Details

gauss

public boolean gauss(float[] img,
 int[] dim,
 double sigma)

Compute Gaussian convolution with and extend-border-pixels outofbounds strategy

Parameters:

img - - the image (1d/2d/3d) as flat float array

dim - - the dimensions of the image in 1d/2d/3d

sigma - - the sigma for all dimensions

Returns:

success? true or false

gauss

public boolean gauss(float[] img,
 int[] dim,
 double[] sigma)

Compute Gaussian convolution with and extend-border-pixels outofbounds strategy

Parameters:

img - - the image (1d/2d/3d) as flat float array

dim - - the dimensions of the image in 1d/2d/3d

sigma - - the sigma per dimension

Returns:

success? true or false

gauss

public boolean gauss(float[] img,
 int[] dim,
 float[] sigma)

gauss

public boolean gauss(float[] img,
 int[] dim,
 double sigma,
 CUDASeparableConvolutionFunctions.OutOfBounds oobs,
 float oobsValue)

Compute Gaussian convolution

Parameters:

img - - the image (1d/2d/3d) as flat float array

dim - - the dimensions of the image in 1d/2d/3d

sigma - - the sigma for all dimensions

oobs - - the OutOfBounds strategy

oobsValue - - value to use in oobs

Returns:

success? true or false

gauss

public boolean gauss(float[] img,
 int[] dim,
 double[] sigma,
 CUDASeparableConvolutionFunctions.OutOfBounds oobs,
 float oobsValue)

Compute Gaussian convolution

Parameters:

img - - the image (1d/2d/3d) as flat float array

dim - - the dimensions of the image in 1d/2d/3d

sigma - - the sigma per dimension

oobs - - the OutOfBounds strategy

oobsValue - - the value for out of image pixels if the OutOfBoundsStrategy is OutOfBounds.VALUE

Returns:

success? true or false

gauss

public static final boolean gauss(float[] img,
 int[] dim,
 double sigma,
 CUDASeparableConvolutionFunctions.OutOfBounds oobs,
 float oobsValue,
 CUDASeparableConvolution cuda,
 int cudaDeviceId)

Compute Gaussian convolution

Parameters:

img - - the image (1d/2d/3d) as flat float array

dim - - the dimensions of the image in 1d/2d/3d

sigma - - the sigma for all dimensions

oobs - - the OutOfBounds strategy

oobsValue - - the value for out of image pixels if the OutOfBoundsStrategy is OutOfBounds.VALUE

cuda - - The CUDASeparableConvolution interface that loaded the external native library

cudaDeviceId - - which CUDA device to use, -1 means single-threaded computation on CPU using native code

Returns:

success? true or false

gauss

public static final boolean gauss(float[] img,
 int[] dim,
 double[] sigma,
 CUDASeparableConvolutionFunctions.OutOfBounds oobs,
 float oobsValue,
 CUDASeparableConvolution cuda,
 int cudaDeviceId)

Compute Gaussian convolution

Parameters:

img - - the image (1d/2d/3d) as flat float array

dim - - the dimensions of the image in 1d/2d/3d

sigma - - the sigma per dimension

oobs - - the OutOfBounds strategy

oobsValue - - the value for out of image pixels if the OutOfBoundsStrategy is OutOfBounds.VALUE

cuda - - The CUDASeparableConvolution interface that loaded the external native library

cudaDeviceId - - which CUDA device to use, -1 means single-threaded computation on CPU using native code

Returns:

success? true or false

getCUDAKernels

public static float[][] getCUDAKernels(double[] sigma,
 int[] supportedKernelSizes)

getFloatKernelPadded

public static float[] getFloatKernelPadded(double[] kernel,
 int size)