MULTIPLE FEATURE IMAGE SEGMENTATION ALGORITHM

Abstract

In this paper we present an algorithm based on the pyramidal approach. For local information we device a new uniformity criterion, called Fisher criterion, which utilizes all multiple perceptual impression information. This uniformity criterion for region classification has been formulated recursively for building the pyramid in a bottom up fashion. Each region in the image is extracted individually by first reaching its seed region and then growing its possible neighbouring areas. The statistics of the multiple features used here are derived form the CIE (L*a*b*) system which is assumed possesses uniform characteristics of the colour perception