SEGMENTATION OF COLOR IMAGES BASED ON THE GRAVITATIONAL CLUSTERING CONCEPT

A new clustering algorithm derived from the Markovian model of the gra vitational clustering concept is proposed that works in the RGB measur ement space for color image. To enable the model to be applicable in i mage segmentation, the new algorithm imposes a clustering constraint a t each clustering iteration to control and determine the formation of multiple clusters, Using such constraint to limit the attraction betwe en clusters, a termination condition can be easily defined. The new cl ustering algorithm is evaluated objectively and subjectively on three different images against the K-means clustering algorithm, the recursi ve histogram clustering algorithm for color (also known as the multi-s pectral thresholding), the Hedley-Yan algorithm, and the widely used s eed-based region growing algorithm. From the evaluation, it is observe d that the new algorithm exhibits the following characteristics: (1) i ts objective measurement figures are comparable with the best in this group of segmentation algorithms; (2) it generates smoother region bou ndaries; (3) the segmented boundaries align closely with the original boundaries; and (4) it forms a meaningful number of segmented regions. (C) 1998 Society of Photo-Optical Instrumentation Engineers. [S0091-3 286(98)02803-7].