at.mug.iqm.img.bundle.jai.stuff

Class FuzzyLICMeansImageClustering

java.lang.Object

java.lang.Thread

at.mug.iqm.img.bundle.jai.stuff.ImageProcessingTask

at.mug.iqm.img.bundle.jai.stuff.FuzzyLICMeansImageClustering

All Implemented Interfaces:

IProgressEmitter, ICancelable, IProcessingTask, Runnable

public class FuzzyLICMeansImageClustering
extends ImageProcessingTask

This class implements a basic Fuzzy C-Means clustering algorithm as an image processing task. This implementation tries to speed things up, but needs to keep all image data on memory. This implementation deals only with integer-like pixel data but can cluster N-dimensional data. This implementation can return ranked images (i.e. second, third, etc. best choices) as well as the membership- function based images.

Changes

• 2010 05 adapted to FLICM Clustering according to Krinidis S, Chatzis V., A Robust Fuzzy Local Information C-Means Clustering Algorithm, IEEE Trans. Img. Proc., 19, 2010, 1328-1337.
• 2010 05 added methods: calculateFuzzyFactorGs(), calcNorm2() adapted methods: calculateMFsFromClusterCenters()

Since:

2009 06 Original from Rafael Santos (rafael.santos@lac.inpe.br), see below

Author:

Helmut Ahammer

Nested Class Summary

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

Thread.State, Thread.UncaughtExceptionHandler

Field Summary

Fields

Modifier and Type	Field and Description
private float[]	aPixel
private float[][]	clusterCenters
private double	epsilon
private float	fuzziness
private float[][][]	fuzzyFactorG
private boolean	hasFinished
private int	height
private int[]	inputData
private int	iteration
private double	j
private int	maxIterations
private float[][][]	membership
private int	numBands
private int	numClusters
private short[][]	outputData
private javax.media.jai.PlanarImage	pInput
private long	position
private int	width

Fields inherited from class at.mug.iqm.img.bundle.jai.stuff.ImageProcessingTask

isCancelable, parentTask, virtual

Fields inherited from class java.lang.Thread

MAX_PRIORITY, MIN_PRIORITY, NORM_PRIORITY

Constructor Summary

Constructors

Constructor and Description
FuzzyLICMeansImageClustering(javax.media.jai.PlanarImage pInput, int numClusters, int maxIterations, float fuzziness, double epsilon) The constructor for the class, which sets the input image, the number of desired clusters, the maximum number of iterations, the fuzziness ("m" value) and a value that will be used to decide whether the convergence has stopped.

Method Summary

Modifier and Type	Method and Description
private float	calcDistance(float[] a1, float[] a2) This method calculates the Euclidean distance between two N-dimensional vectors.
private float	calcNormP(float[] vec, float p) added by Ahammer 2010 05 This method calculates the p-norm.
private float	calcSquaredDistance(float[] a1, float[] a2) This method calculates the squared distance between two N-dimensional vectors.
private void	calculateClusterCentersFromMFs() This method calculates the cluster centers from the membership functions.
private void	calculateFuzzyFactorGs() added by Ahammer 2010 05 This method calculates the fuzzy factor G
private void	calculateMFsFromClusterCenters() adapted by Ahammer 2010 05 This method calculates the membership functions from the cluster centers.
private double	calculateObjectiveFunction()
float[][]	getClusterCenters() This method returns the cluster centers.
double	getCompactnessAndSeparation() This method returns the Compactness and Separation measure of cluster validity (see Fuzzy Algorithms With Applications to Image Processing and Pattern Recognition, Zheru Chi, Hong Yan, Tuan Pham, World Scientific, pp.
double	getPartitionCoefficient() This method returns the Partition Coefficient measure of cluster validity (see Fuzzy Algorithms With Applications to Image Processing and Pattern Recognition, Zheru Chi, Hong Yan, Tuan Pham, World Scientific, pp.
double	getPartitionEntropy() This method returns the Partition Entropy measure of cluster validity (see Fuzzy Algorithms With Applications to Image Processing and Pattern Recognition, Zheru Chi, Hong Yan, Tuan Pham, World Scientific, pp.
long	getPosition() This method returns a measure of the progress of the algorithm.
javax.media.jai.TiledImage	getRankedImage(int rank) This method will return a rank image, i.e.
private int	getRankedIndex(float[] data, int rank) This method returns the ranked index of a cluster from an array containing the membership functions.
private float	getRankedMF(float[] data, int rank) This method returns the ranked membership function of a cluster from an array containing the membership functions.
javax.media.jai.TiledImage	getRankedMFImage(int rank) This method will return a membership function image, i.e.
long	getSize() This method returns the estimated size (steps) for this task.
boolean	isFinished() This method returns true if the clustering has finished.
void	run() This method performs the bulk of the processing.

Methods inherited from class at.mug.iqm.img.bundle.jai.stuff.ImageProcessingTask

addProgressListener, fireProgressChanged, getParentTask, getPcs, getProgressListeners, isCancelable, isCancelled, isVirtual, setCancelable, setParentTask, setPcs, setVirtual

Methods inherited from class java.lang.Thread

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

Methods inherited from class java.lang.Object

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

Field Detail

pInput

private javax.media.jai.PlanarImage pInput

width

private int width

height

private int height

numBands

private int numBands

maxIterations

private int maxIterations

numClusters

private int numClusters

fuzziness

private float fuzziness

membership

private float[][][] membership

fuzzyFactorG

private float[][][] fuzzyFactorG

iteration

private int iteration

j

private double j

epsilon

private double epsilon

hasFinished

private boolean hasFinished

position

private long position

clusterCenters

private float[][] clusterCenters

inputData

private int[] inputData

aPixel

private float[] aPixel

outputData

private short[][] outputData

Constructor Detail

FuzzyLICMeansImageClustering

public FuzzyLICMeansImageClustering(javax.media.jai.PlanarImage pInput,
                                    int numClusters,
                                    int maxIterations,
                                    float fuzziness,
                                    double epsilon)

The constructor for the class, which sets the input image, the number of desired clusters, the maximum number of iterations, the fuzziness ("m" value) and a value that will be used to decide whether the convergence has stopped. It also allocates the required memory.

Parameters:

pInput - the input planar image.

numClusters - the desired number of clusters.

maxIterations - the maximum number of iterations.

fuzziness - the fuzziness (a.k.a. the "m" value)

epsilon - a small value used to verify if clustering has converged.

Method Detail

run

public void run()

This method performs the bulk of the processing. It runs the classic Fuzzy C-Means clustering algorithm: 1 - Calculate the cluster centers. 2 - Update the membership function. 3 - Calculate statistics and repeat from 1 if needed.

Specified by:

run in interface Runnable

Specified by:

run in class ImageProcessingTask

calculateClusterCentersFromMFs

private void calculateClusterCentersFromMFs()

This method calculates the cluster centers from the membership functions.

calculateMFsFromClusterCenters

private void calculateMFsFromClusterCenters()

adapted by Ahammer 2010 05 This method calculates the membership functions from the cluster centers.

calculateFuzzyFactorGs

private void calculateFuzzyFactorGs()

added by Ahammer 2010 05 This method calculates the fuzzy factor G

calculateObjectiveFunction

private double calculateObjectiveFunction()

calcDistance

private float calcDistance(float[] a1,
                           float[] a2)

This method calculates the Euclidean distance between two N-dimensional vectors.

Parameters:

a1 - the first data vector.

a2 - the second data vector.

Returns:

the Euclidean distance between those vectors.

calcNormP

private float calcNormP(float[] vec,
                        float p)

added by Ahammer 2010 05 This method calculates the p-norm.

Parameters:

vec - data vector..

Returns:

the p-norm of vector.

getSize

public long getSize()

This method returns the estimated size (steps) for this task. The value is, of course, an approximation, just so we will be able to give the user a feedback on the processing time. In this case, the value is calculated as the number of loops in the run() method.

Specified by:

getSize in class ImageProcessingTask

Returns:

the size of the image processing size.

getPosition

public long getPosition()

This method returns a measure of the progress of the algorithm.

Specified by:

getPosition in class ImageProcessingTask

Returns:

the position of the task processing.

isFinished

public boolean isFinished()

This method returns true if the clustering has finished.

Specified by:

isFinished in class ImageProcessingTask

Returns:

true if the processing task has finished.

getRankedImage

public javax.media.jai.TiledImage getRankedImage(int rank)

This method will return a rank image, i.e. an image which pixels are the cluster centers of the Nth best choice for the classification. For example, if the membership functions for a pixel of an image clustered with six clusters are [0.10 0.25 0.40 0.20 0.03 0.02] and we ask for the image with rank 2 (ranks starts with zero), for that pixel the third best choice for cluster index will be selected (0.20) and the centers of the cluster with index 3 will be used. It is important to notice that this method can be called while the clustering task (method run) is running, and the resulting image is not guaranteed to be close to a clustered result.

Parameters:

rank - the desired rank for the classification.

Returns:

a TiledImage with the classification results considering that rank.

getRankedMFImage

public javax.media.jai.TiledImage getRankedMFImage(int rank)

This method will return a membership function image, i.e. an image which pixels correspond to the membership functions for the cluster which is the Nth best choice for the classification. For example, if the membership functions for a pixel of an image clustered with six clusters are [0.10 0.25 0.40 0.20 0.03 0.02] and we ask for the membership function image with rank 2 (ranks starts with zero), for that pixel the third best membership function will be selected (0.20) and used (scaled by 255) as the pixel value. It is important to notice that this method can be called while the clustering task (method run) is running, and the resulting image is not guaranteed to be close to a clustered result.

Parameters:

rank - the desired rank for the classification.

Returns:

a TiledImage with the membership function value results considering that rank.

getRankedIndex

private int getRankedIndex(float[] data,
                           int rank)

This method returns the ranked index of a cluster from an array containing the membership functions. For example, if the array contains [0.10 0.25 0.40 0.20 0.03 0.02] and we ask for index with rank 2 (ranks starts with zero), the third best choice will be selected (0.20) and its index (3) will be returned.

Parameters:

data - the array with the membership functions.

rank - the rank of the cluster we want to get.

Returns:

the index of the cluster.

getRankedMF

private float getRankedMF(float[] data,
                          int rank)

This method returns the ranked membership function of a cluster from an array containing the membership functions. For example, if the array contains [0.10 0.25 0.40 0.20 0.03 0.02] and we ask for the MF with rank 2 (ranks starts with zero), the third best choice will be selected (0.20) and returned.

Parameters:

data - the array with the membership functions.

rank - the rank of the cluster we want to get.

Returns:

the MF with that rank.

getPartitionCoefficient

public double getPartitionCoefficient()

This method returns the Partition Coefficient measure of cluster validity (see Fuzzy Algorithms With Applications to Image Processing and Pattern Recognition, Zheru Chi, Hong Yan, Tuan Pham, World Scientific, pp. 91)

getPartitionEntropy

public double getPartitionEntropy()

This method returns the Partition Entropy measure of cluster validity (see Fuzzy Algorithms With Applications to Image Processing and Pattern Recognition, Zheru Chi, Hong Yan, Tuan Pham, World Scientific, pp. 91)

getCompactnessAndSeparation

public double getCompactnessAndSeparation()

This method returns the Compactness and Separation measure of cluster validity (see Fuzzy Algorithms With Applications to Image Processing and Pattern Recognition, Zheru Chi, Hong Yan, Tuan Pham, World Scientific, pp. 93)

calcSquaredDistance

private float calcSquaredDistance(float[] a1,
                                  float[] a2)

This method calculates the squared distance between two N-dimensional vectors.

Parameters:

a1 - the first data vector.

a2 - the second data vector.

Returns:

the squared distance between those vectors.

getClusterCenters

public float[][] getClusterCenters()

This method returns the cluster centers. added 2009 08

Returns:

cluster centers