net.sf.ij_plugins.clustering

Class KMeans

java.lang.Object

net.sf.ij_plugins.clustering.KMeans2D

public final class KMeans
extends java.lang.Object

Pixel-based multi-band image segmentation using k-means clustering algorithm.

k-means clustering performs pixel-based segmentation of multi-band images. An image stack is interpreted as a set of bands corresponding to the same image. For instance, an RGB color images has three bands: red, green, and blue. Each pixels is represented by an n-valued vector , where n is a number of bands, for instance, a 3-value vector [r,g,b] in case of a color image. Each cluster is defined by its centroid in n-dimensional space. Pixels are grouped by their proximity to cluster's centroids. Cluster centroids are determined using a heuristics: initially centroids are randomly initialized and then their location is interactively optimized. For more information on this and other clustering approaches see: Anil K. Jain and Richard C. Dubes, Algorithms for Clustering Data, Prentice Hall, 1988. http://homepages.inf.ed.ac.uk/rbf/BOOKS/JAIN/Clustering_Jain_Dubes.pdf

Author:

Jarek Sacha

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	KMeans.Config Configurable parameters of the k-means algorithm.

Constructor Summary

Constructors

Constructor and Description
KMeans()
KMeans(KMeans.Config config)

Method Summary

Methods

Modifier and Type	Method and Description
int	closestCluster(float[] x) Find index of the cluster closest to the sample x.
ij.ImageStack	getCentroidValueImage() Returns stack where discovered clusters can be represented by replacing pixel values in a cluster by the value of the centroid of that cluster.
ij.ImageStack	getClusterAnimation() Return stack representing clustering optimization.
float[][]	getClusterCenters() Return location of cluster centers.
long	getNumberOfStepsToConvergence()
ij.process.ByteProcessor	run(ij.ImageStack stack) Perform k-means clustering of the input stack.

Methods inherited from class java.lang.Object

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

Constructor Detail

KMeans

public KMeans()

KMeans

public KMeans(KMeans.Config config)

Method Detail

run

public ij.process.ByteProcessor run(ij.ImageStack stack)

Perform k-means clustering of the input stack. Elements of the stack must be of type FloatProcessor.

Parameters:

stack - stack representing a multi-band image.

Returns:

segmented image.

getClusterCenters

public float[][] getClusterCenters()

Return location of cluster centers.

Returns:

array of cluster centers. First index refers to cluster number.

getClusterAnimation

public ij.ImageStack getClusterAnimation()

Return stack representing clustering optimization. This will return not null value only when configuration parameters clusterAnimationEnabled is set to true.

Returns:

stack representing cluster optimization, can return null.

getCentroidValueImage

public ij.ImageStack getCentroidValueImage()

Returns stack where discovered clusters can be represented by replacing pixel values in a cluster by the value of the centroid of that cluster.

Returns:

centroid value image

getNumberOfStepsToConvergence

public long getNumberOfStepsToConvergence()

closestCluster

public int closestCluster(float[] x)

Find index of the cluster closest to the sample x. run(ij.ImageStack) must be run before calling this method.

Parameters:

x - test point, number of values must be the same is input stack size.

Returns:

index of the closest cluster.

See Also:

run(ij.ImageStack)