Package gov.nih.mipav.model.algorithms.filters

Class AlgorithmCoherenceEnhancingDiffusion

java.lang.Object

java.lang.Thread

gov.nih.mipav.model.algorithms.AlgorithmBase

gov.nih.mipav.model.algorithms.filters.AlgorithmCoherenceEnhancingDiffusion

All Implemented Interfaces:

java.awt.event.ActionListener, java.awt.event.WindowListener, java.lang.Runnable, java.util.EventListener

public class AlgorithmCoherenceEnhancingDiffusion
extends AlgorithmBase

Algorithm to apply Coherence Enhancing Anisotropic Diffusion

For color form a common structure tensor for all colors by averaging the structure tensors for each color.

References:

1. Joachim Weickert
   Chapter 15 Nonlinear Diffusion Filtering in Handbook of Computer Vision and Applications, Volume 2 Signal Processing and Pattern Recognition. The book is published by Academic Press, ISBN 0-12-379772-1
2. Algorithms for Non-Linear Diffusion Matlab in a Literate Programming Style by Rein van den Boomgaard, Intelligent Sensory Information Systems, University of Amsterdam, The Netherlands http://carol.wins.uva.nl/~rein/nldiffusionweb/material.html
3. Coherence-Enhancing Diffusion of Colour Images by Joachim Weickert, A. Sanfeliu, J.J. Villanueva, J. Vitria, editors, Proc. VII National Symposium on Pattern Recognition and Image Analysis (VII NSPRIA, Barcelona, April 21-25, 1997), Vol. 1, pp. 239-244, 1997.
4. Coherence-Enhancing Diffusion of Colour Images by Joachim Weickert, Image and Vision Computing, Vol. 17, 1999, pp. 201-212.
5. Coherence-Enhancing Diffusion Filtering by Joachim Weickert, International Journal of Computer Vision, Vol. 31, No. 2/3, pp. 111-127, April, 1999

Version:

1.0; 31, July 2003

Author:

Paul F. Hemler, Ph.D.

Nested Class Summary

Nested classes/interfaces inherited from class java.lang.Thread

java.lang.Thread.State, java.lang.Thread.UncaughtExceptionHandler

Field Summary

Fields

Modifier and Type	Field	Description
(package private) float	derivativeScale	Gaussian scale for derivative operations.
(package private) boolean	do25D	flag indicating 2.5D processing.
private float[]	dxKernelX	Kernels for the X derivative.
private float[]	dxKernelY	DOCUMENT ME!
private float[]	dxKernelZ	DOCUMENT ME!
private float[]	dyKernelX	Kernels for the Y derivative.
private float[]	dyKernelY	DOCUMENT ME!
private float[]	dyKernelZ	DOCUMENT ME!
private float[]	dzKernelX	Kernels for the Z derivative.
private float[]	dzKernelY	DOCUMENT ME!
private float[]	dzKernelZ	DOCUMENT ME!
private WildMagic.LibFoundation.NumericalAnalysis.Eigenf	eigenSystemAlgo	Eigensystem solver.
private boolean	entireImage	Whether to process the entire image, or just the portion within the currently selected VOI.
(package private) float	gaussianScale	Gaussian scale for structure tensor smoothing.
private float[]	intermediateBuffer	Reference to intermediate buffer.
(package private) double	k	diffusitivity denominator.
(package private) double	k2	diffusitivity denominator squared.
private int	kRadius	Width of the derivative buffer.
(package private) int	numIterations	The number of iterations.
private float[]	skernelX	Kernels for the Smoothing kernel.
private float[]	skernelY	DOCUMENT ME!
private float[]	skernelZ	DOCUMENT ME!
private int	skRadius	Width of the smoothing buffer.
(package private) float	timeStep	The time parameter step size, called tau in the algorithm description In 2D for stability the time step should be <= 0.25.
private int	xDim	image dimensions.
private int	yDim	image dimensions.
private int	zDim	image dimensions.

Fields inherited from class gov.nih.mipav.model.algorithms.AlgorithmBase

destFlag, destImage, image25D, mask, maxProgressValue, minProgressValue, multiThreadingEnabled, nthreads, progress, progressModulus, progressStep, runningInSeparateThread, separable, srcImage, threadStopped

Fields inherited from class java.lang.Thread

MAX_PRIORITY, MIN_PRIORITY, NORM_PRIORITY

Constructor Summary

Constructors

Constructor	Description
AlgorithmCoherenceEnhancingDiffusion(ModelImage srcImg, int numI, float k, float derivativeScale, float gaussianScale, boolean do25D, boolean entireImage)	Creates a new AlgorithmCoherenceEnhancingDiffusion object.
AlgorithmCoherenceEnhancingDiffusion(ModelImage destImg, ModelImage srcImg, int numI, float k, float derivativeScale, float gaussianScale, boolean do25D, boolean entireImage)	Creates a new AlgorithmCoherenceEnhancingDiffusion object.

Method Summary

Modifier and Type	Method	Description
private void	computeDx(float[] dx, float[] src)	DOCUMENT ME!
private void	computeDx3D(float[] dx, float[] src)	DOCUMENT ME!
private void	computeDy(float[] dy, float[] src)	DOCUMENT ME!
private void	computeDy3D(float[] dy, float[] src)	DOCUMENT ME!
private void	computeDz(float[] dz, float[] src)	DOCUMENT ME!
void	finalize()	Prepares this class for destruction.
private void	gaussianSmooth(float[] gs, float[] src)	DOCUMENT ME!
private void	gaussianSmooth3D(float[] gs, float[] src)	DOCUMENT ME!
private float	getVal(float[] buffer, int col, int row)	Returns the value in a 2D image buffer of the pixel at location row, col.
private float	getVal(float[] buffer, int col, int row, int slice)	Returns the value in a 3D image buffer of the pixel at location slice, row, col.
private void	init()	Initialize buffers and kernels.
private static float	max(float a, float b)	Computes the maximum of two floating point numbers passed in.
private void	run2D(int numImages)	Starts the algorithm for 2D images.
private void	run2DC(int numImages)	Starts the algorithm for 2D images.
private void	run3D()	Starts the algorithm for 3D images.
private void	run3DC()	Starts the algorithm for 3D color images.
void	runAlgorithm()	Starts the program.
private void	setVal(float val, float[] buffer, int col, int row)	Sets the value in a 2D image buffer at location row, col.
private void	setVal(float val, float[] buffer, int col, int row, int slice)	Sets the value in a 3D image buffer at location row, col.

Methods inherited from class gov.nih.mipav.model.algorithms.AlgorithmBase

actionPerformed, addListener, addProgressChangeListener, calculateImageSize, calculatePrincipleAxis, computeElapsedTime, computeElapsedTime, convertIntoFloat, delinkProgressToAlgorithm, delinkProgressToAlgorithmMulti, displayError, errorCleanUp, fireProgressStateChanged, fireProgressStateChanged, fireProgressStateChanged, fireProgressStateChanged, fireProgressStateChanged, generateProgressValues, getDestImage, getElapsedTime, getMask, getMaxProgressValue, getMinProgressValue, getNumberOfThreads, getProgress, getProgressChangeListener, getProgressChangeListeners, getProgressModulus, getProgressStep, getProgressValues, getSrcImage, isCompleted, isImage25D, isMultiThreadingEnabled, isRunningInSeparateThread, isThreadStopped, linkProgressToAlgorithm, linkProgressToAlgorithm, makeProgress, notifyListeners, removeListener, removeProgressChangeListener, run, setCompleted, setImage25D, setMask, setMaxProgressValue, setMinProgressValue, setMultiThreadingEnabled, setNumberOfThreads, setProgress, setProgressModulus, setProgressStep, setProgressValues, setProgressValues, setRunningInSeparateThread, setSrcImage, setStartTime, setThreadStopped, startMethod, windowActivated, windowClosed, windowClosing, windowDeactivated, windowDeiconified, windowIconified, windowOpened

Methods inherited from class java.lang.Thread

activeCount, checkAccess, clone, countStackFrames, currentThread, dumpStack, enumerate, getAllStackTraces, getContextClassLoader, getDefaultUncaughtExceptionHandler, getId, getName, getPriority, getStackTrace, getState, getThreadGroup, getUncaughtExceptionHandler, holdsLock, interrupt, interrupted, isAlive, isDaemon, isInterrupted, join, join, join, onSpinWait, resume, setContextClassLoader, setDaemon, setDefaultUncaughtExceptionHandler, setName, setPriority, setUncaughtExceptionHandler, sleep, sleep, start, stop, suspend, toString, yield

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

derivativeScale

float derivativeScale

Gaussian scale for derivative operations.

do25D

boolean do25D

flag indicating 2.5D processing.

gaussianScale

float gaussianScale

Gaussian scale for structure tensor smoothing.

k

double k

diffusitivity denominator.

k2

double k2

diffusitivity denominator squared.

numIterations

int numIterations

The number of iterations.

timeStep

float timeStep

The time parameter step size, called tau in the algorithm description In 2D for stability the time step should be <= 0.25. In 3D for stability the time step should be <= 1/6.

dxKernelX

private float[] dxKernelX

Kernels for the X derivative.

dxKernelY

private float[] dxKernelY

DOCUMENT ME!

dxKernelZ

private float[] dxKernelZ

DOCUMENT ME!

dyKernelX

private float[] dyKernelX

Kernels for the Y derivative.

dyKernelY

private float[] dyKernelY

DOCUMENT ME!

dyKernelZ

private float[] dyKernelZ

DOCUMENT ME!

dzKernelX

private float[] dzKernelX

Kernels for the Z derivative.

dzKernelY

private float[] dzKernelY

DOCUMENT ME!

dzKernelZ

private float[] dzKernelZ

DOCUMENT ME!

eigenSystemAlgo

private WildMagic.LibFoundation.NumericalAnalysis.Eigenf eigenSystemAlgo

Eigensystem solver.

entireImage

private boolean entireImage

Whether to process the entire image, or just the portion within the currently selected VOI.

intermediateBuffer

private float[] intermediateBuffer

Reference to intermediate buffer.

kRadius

private int kRadius

Width of the derivative buffer.

skernelX

private float[] skernelX

Kernels for the Smoothing kernel.

skernelY

private float[] skernelY

DOCUMENT ME!

skernelZ

private float[] skernelZ

DOCUMENT ME!

skRadius

private int skRadius

Width of the smoothing buffer.

xDim

private int xDim

image dimensions.

yDim

private int yDim

image dimensions.

zDim

private int zDim

image dimensions.

Constructor Detail

AlgorithmCoherenceEnhancingDiffusion

public AlgorithmCoherenceEnhancingDiffusion(ModelImage srcImg,
                                            int numI,
                                            float k,
                                            float derivativeScale,
                                            float gaussianScale,
                                            boolean do25D,
                                            boolean entireImage)

Creates a new AlgorithmCoherenceEnhancingDiffusion object.

Parameters:

srcImg - Source image model

numI - Number of iterations

k - Diffusitivity denominator

derivativeScale - Scale of the gaussian when taking derivatives

gaussianScale - Scale of the gaussian for smoothing the structure tensor

do25D - Flag indicating 2.5D processing

entireImage - whether to filter the entire image or just the VOI region

AlgorithmCoherenceEnhancingDiffusion

public AlgorithmCoherenceEnhancingDiffusion(ModelImage destImg,
                                            ModelImage srcImg,
                                            int numI,
                                            float k,
                                            float derivativeScale,
                                            float gaussianScale,
                                            boolean do25D,
                                            boolean entireImage)

Creates a new AlgorithmCoherenceEnhancingDiffusion object.

Parameters:

destImg - Image model where result image is to stored

srcImg - Source image model

numI - Number of iterations

k - Diffusitivity denominator

derivativeScale - Scale of the gaussian when taking derivatives

gaussianScale - Scale of the gaussian for smoothing the structure tensor

do25D - Flag indicating 2.5D processing

entireImage - whether to filter the entire image or just the VOI region

Method Detail

finalize

public void finalize()

Prepares this class for destruction.

Overrides:

finalize in class AlgorithmBase

runAlgorithm

public void runAlgorithm()

Starts the program.

Specified by:

runAlgorithm in class AlgorithmBase

max

private static float max(float a,
                         float b)

Computes the maximum of two floating point numbers passed in.

Parameters:

a - float The first value

b - float The second value

Returns:

float The maximum of values a and b

computeDx

private void computeDx(float[] dx,
                       float[] src)

DOCUMENT ME!

Parameters:

dx - float [] destination buffer

src - float [] source buffer

Member function computes the derivative of the source buffer in the X-direction using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D convolution with the derivative of a Gaussian kernel in the X-direction and stores the result in an intermediate buffer. Once the intermediate buffer is completely filled, a smoothing in the Y-direction is performed using a 1D Gaussian kernel computed as in the MATLAB paper.

computeDx3D

private void computeDx3D(float[] dx,
                         float[] src)

DOCUMENT ME!

Parameters:

dx - float [] destination buffer

src - float [] source buffer

Member function computes the derivative of the source buffer in the X-direction using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D convolution with the derivative of a Gaussian kernel in the X-direction and stores the result in dx buffer. Once dx is completely filled, a smoothing in the Y-direction is performed using a 1D Gaussian kernel and the intermediate result is stored in intermediateBuffer. Once intermediateBuffer is completely filled, a smoothing in the Z-direction is performed and the final result is placed in dx.

computeDy

private void computeDy(float[] dy,
                       float[] src)

DOCUMENT ME!

Parameters:

dy - float [] destination buffer

src - float [] source buffer

Member function computes the derivative of the source buffer in the Y-direction using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D convolution with a Gaussian kernel in the X-direction and stores the result in an intermediate buffer. Once the intermediate buffer is completely filled, a 1D convolution in the Y-direction is performed using a 1D Gaussian derivative kernel computed as in the MATLAB paper.

computeDy3D

private void computeDy3D(float[] dy,
                         float[] src)

DOCUMENT ME!

Parameters:

dy - float [] destination buffer

src - float [] source buffer

Member function computes the derivative of the source buffer in the Y-direction using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D smoothing convolution with a Gaussian kernel in the X-direction and stores the result in dy buffer. Once dy is completely filled, a convolution in the Y-direction is performed using the derivative of a 1D Gaussian kernel and the intermediate result is stored in intermediateBuffer. Once intermediateBuffer is completely filled, a smoothing in the Z-direction is performed and the final result is placed in dy.

computeDz

private void computeDz(float[] dz,
                       float[] src)

DOCUMENT ME!

Parameters:

dz - float [] destination buffer

src - float [] source buffer

Member function computes the derivative of the source buffer in the Z-direction using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D convolution with a Gaussian kernel in the X-direction and stores the result in dz buffer. Once dz is completely filled, a smoothing in the Y-direction is performed using a 1D Gaussian kernel and the intermediate result is stored in intermediateBuffer. Once intermediateBuffer is completely filled, a convolution with the derivative of a Gaussian kernel in the Z-direction is performed and the final result is placed in dz.

gaussianSmooth

private void gaussianSmooth(float[] gs,
                            float[] src)

DOCUMENT ME!

Parameters:

gs - float [] destination buffer

src - float [] source buffer

Member function gaussian smoothes the source buffer in the x and y directions using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D convolution with a Gaussian kernel in the X-direction and stores the result in an intermediate buffer. Once the intermediate buffer is completely filled, a 1D convolution in the Y-direction is performed using a 1D Gaussian kernel.

gaussianSmooth3D

private void gaussianSmooth3D(float[] gs,
                              float[] src)

DOCUMENT ME!

Parameters:

gs - float [] destination buffer

src - float [] source buffer

Member function gaussian smoothes the source buffer in the x,y, and z directions using the kernel formulation as specified in the MATLAB paper. The function uses seperable convolution, which means it first does a 1D convolution with a Gaussian kernel in the X-direction and stores the result in gs. Once the gs buffer is completely filled, a 1D convolution in the Y-direction is performed using a 1D Gaussian kernel and the result is stored in intermediateBuffer. Once intermediateBuffer is completely filled, a 1D convolution in the Z-direction is performed using a 1D Gaussian kernel and the result is placed in gs.

getVal

private float getVal(float[] buffer,
                     int col,
                     int row)

Returns the value in a 2D image buffer of the pixel at location row, col.

Parameters:

buffer - float[] The image buffer

col - int Column index

row - int Row index

Returns:

float The pixel value

getVal

private float getVal(float[] buffer,
                     int col,
                     int row,
                     int slice)

Returns the value in a 3D image buffer of the pixel at location slice, row, col.

Parameters:

buffer - float[] The image buffer

col - int Column index

row - int Row index

slice - int Slice index

Returns:

float The pixel value

init

private void init()

Initialize buffers and kernels.

run2D

private void run2D(int numImages)

Starts the algorithm for 2D images.

Parameters:

numImages - int The number of 2D images

run2DC

private void run2DC(int numImages)

Starts the algorithm for 2D images.

Parameters:

numImages - int The number of 2D images

run3D

private void run3D()

Starts the algorithm for 3D images.

run3DC

private void run3DC()

Starts the algorithm for 3D color images.

setVal

private void setVal(float val,
                    float[] buffer,
                    int col,
                    int row)

Sets the value in a 2D image buffer at location row, col.

Parameters:

val - float The value to set the pixel to

buffer - float[] 2D image buffer

col - int Column index

row - int Row index

setVal

private void setVal(float val,
                    float[] buffer,
                    int col,
                    int row,
                    int slice)

Sets the value in a 3D image buffer at location row, col.

Parameters:

val - float The value to set the pixel to

buffer - float[] 3D image buffer

col - int Column index

row - int Row index

slice - int Slice index