Adaptive image restoration using Hopfield neural network
Souheila Ghennam, K. Benmahammed · 2002
In this paper, a solution to the restoration problem of degraded and noisy image is proposed, using the modified Hopfield network. To improve the network performances, the eliminating highest error EHE is used. In order to keep the image structures unaltered, an adaptive regularization scheme is employed that allows better compromise between inversion degradation process and smoothing. To do this, the statistics analysis is used to assign each pixel one regularization parameter according to its spatial activity.