A Partitioned Modified Hopfield Neural Network Algorithm for Real-Time Image Restoration

In recent years attention has been turned to the use of neural network-derived algorithms to restore images using a model-based approach. Considering an M by M input image, in most cases the image degradation model is a spatially and temporally invariant linear distortion described by the equation (Pratt [1]) (Andrews and Hunt [2]): Where f and g are the M2 by 1 lexicographically organized original and degraded resultant image vectors respectively, H is a matrix operator which gives the same result as a convolution of the original image with the distortion point spread function, and n is an additive noise vector. Let L be equal to M2.