PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Image thresholding by using statistics for determining the number of clusters
Jinghao Xue and Mike Titterington
Pattern Recognition Letters 2010.

Abstract

This paper aims to demonstrate that some statistics originally developed in cluster analysis for determining the number of clusters can be instead appropriately used in image thresholding to select optimal thresholds. Three highly-cited statistics, namely the Calinski-Harabasz index, the Gap statistic and the Silhouette widths, are chosen and applied to real images and simulated data, for illustrative purposes. The relationship between these statistics and a commonly-used image-thresholding method, Otsu's method, is also briefly discussed.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Article
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Theory & Algorithms
ID Code:7350
Deposited By:Mike Titterington
Deposited On:17 March 2011