Perceptual Segmentation of Nonhomogeneous Dot Patterns

M. Tuceryan and Narendra Ahuja

Abstract

The problem of perceptual segmentation of a dot pattern containing varying density clusters is considered. Geometrical features of the Voronoi neighbourhoods of points are used as similarity measures to group points. A probabilistic relaxation labeling algorithm is designed to label each point as an interior or an edge point of a cluster, based upon the compactness and eccentricity measures of the Voronoi polygons. Experimental results are presented.