Complexation of Radio Images with the Use of a dynamic Maximum Entropy Neural Network

摘要

A new approach is proposed to the problem of improving the quality of radio images (Ris) of exten- sive objects by the complexation of separate images formed simultaneously be several passive remote-sensing (RS) systems. The complexation problem is solved as a nonlinear inverse problem of restoring the true image of an object by the combined processing of various distorted radio images. The minimum of the a priori knowl- edge available is formalized by the maximum entropy (ME) model of the restored image. The solution of the problem is constructed by the minimization of the energy function of a dynamic neural network (NN) with a multilevel vector of states whose parameters integrate the measurement and model information available.