AUTOMATIC SEARCHING FOR CLUSTER CENTERS IN THE PROBLEMS OF GREY IMAGECLUSTERING

The algorithm of clustering that transforms the feature space into the feature matrix aggregate according to the intensity is considered. Ea ch matrix element is equal to the average value of the feature for res pective intensity. The population of vectors forming the matrix is the basis for an automatic searching for cluster centers by the Gauss-Zid el method according to the criterion of an intragroup dispersion minim um or intergroup scatter maximum.