The problem of image segmentation using brightness and color features including originally developed methods as well as methods of segments description applicable for future classification and comparison are reviewing in this paper. The problem of image segmentation is one of important parts in most image processing and recognition applications. Whole system performance usually depends on the segmentation process quality. This paper consists of some descriptions of original and enhanced methodics of image segmentation (using fixed threshold or dynamically changing criteria; iterational algorithms) in compare with wide- known segmentation algorithms. Another issue of this paper is to define a set of segment features allowing us to create computationally low-cost and effective methods of patches comparison and classification. Experimental processing system PDem have been created to investigate problems related with brightness and color criteria in the patches segmentation process and segmentation descripting. All methods are realized in practice.