Optimization of Piecewise Constant Approximation for Segmented Image

In this paper а problem of segmentation of the color image, approached by piecewise constant approximations, is analyzed. The quality of the optimization is estimated by the classical standard deviation of image pixels from the pixels of approximations. The modern versions of the classical methods of image simulating by piecewise constant approximations characterized by minimal values of standard deviation or total squared error are detailed. Four main operations over pixel clusters and appropriate working criterions for the optimized approximation generating are discussed. The algorithmic versions of approximation transformation, providing the enhancement of approximation by standard deviation and also by visual perception for the given number of segments are proposed.