The structural-color analysis of graphic images on the basis of dynamic extent models of cluster

In article the problem of development of a method of the analysis of spatial structure of the color images, providing detection of structural elements and the simultaneous account of their spatial and color context is considered. Parameters of a color graphic image are the spatial - structural parameters determined at centroid transformation: dissipation, weight, values of centroid, and values color component RGB, established at color-tone transformation. Cluster, formed by disorder of characteristics of values of these parameters, are used for recognition of structural-color characteristics of elements of the image. The description of cluster as an ex-awning covering sample of realizations is offered. Proceeding from the analysis of a relative positioning of the various size extent of cluster under diagrams in space of parameters, the deciding rule of definition of an accessory of value to this or that cluster is established due to definition of affinity to the centers of clusters, the sizes of an ex-awning normalized on values.