The use of neural networks in maximum entropy image restoration

The computational properties of analog e1ectrical circuits, recently proposed as models for highly-interconnected networks of non-linear analog neurons, have been found attractive in solving optimization problems. The computation power of these circuits is based on the high connectivity typical of neural systems and on the convergence speed of analog electric circuits in reaching stable states. In this paper we consider the problem of image restoration using the Maximum Entropy (ME) method. This problem is rigorously formulated as a constrained optimization problem in a primal-dual Lagrange multipliers framework. In this way an iterative scheme which can be used to solve the ME image restoration problem can be evidenced. The iterative scheme takes advantage of the computation power of neural networks in solving optimization problems. A possible electric circuit for the neural networks is proposed; some features which can be useful far a practical implementation of the proposed circuit are also reported.