net.sourceforge.javaocr.plugin.cluster
Interface Cluster

All Superinterfaces:

Metric

All Known Implementing Classes:

AbstractBaseCluster, EuclidianDistanceCluster, MahalanobisDistanceCluster, NormalDistributionCluster, SigmaWeightedEuclidianDistanceCluster

public interface Cluster
extends Metric

cluster represents some feature vectors belonging together. cluster can be trained with sampler

Author:

Konstantin Pribluda

Method Summary

double[]	center() centroid of cluster
double	radius(List<double[]> samples) computes maximal distance of sample from center of cluster
void	train(double[] features) train cluster with feature vector

Methods inherited from interface net.sourceforge.javaocr.cluster.Metric

distance, getDimensions

Method Detail

center

double[] center()

centroid of cluster

Returns:

centroid of cluster

train

void train(double[] features)

train cluster with feature vector

Parameters:

features -

radius

double radius(List<double[]> samples)

computes maximal distance of sample from center of cluster

Parameters:

samples - sample group, sample size shall correspond to cluster dimensions