Restoration method using a neural network model

Considers the problem of image restoration degraded by a shift-invariant blur function and corrupted by white Gaussian noise. We propose a modified Hopfield neural network-based image restoration. Two algorithms with two updating modes using the modified Hopfield neural network are presented: (1) sequential updates, and (2) n-simultaneous updates. In the sequential algorithm, only one element of the state is updated at time (t+1), while the rest are left unchanged. In the n-simultaneous algorithm, all elements of the state are updated simultaneously. Lastly, we present some image restoration results which attest to the efficiency of our method.