QUESTIONS ASSOCIATED WITH THE CONSTRUCTION OF NONLINEAR STATISTICAL IMAGE-PROCESSING ALGORITHMS

Classes of nonlinear inertialess statistical transformations of optica l images are considered, the specific details of which are determined by the kernels, which are either the base of the operator in exponenti al form or have an independent value for the transformations. The orig in of the kernels is associated with the laws governing visual percept ion and with the possibilities of ensuring constancy or stabilization of the characteristics of the transformed images, as well as with the optimization of the information in terms of the maximal principle.