Estimation Techniques in Image Restoration - A Survey Approach

Abstract

In the present paper a comparative study of various estimation techniques based on neural network, MATLAB, partial differential equation (PDE) and other proposed models for image restoration are been discussed. An image may be distorted, noisy or blurred and not suitable for extracting desired information or data, so it needs to be restored for desired application. Image restoration techniques are oriented towards modeling the degradation, blur and noise and applying an inverse procedure to reconstruct the original scene. Some techniques require prior estimation of degradation causes and some are based on neural network based training set. For the dependence on various factors and procedures, these techniques require a comparative evaluation.

Keywords: Image restoration, image deblurring, MATLAB, hough transform, markov random field(MRF), multi-valued neuron(MVN), modified hopfield neural network(MHNN), LAB & RBF neural network, PSNR(peak signal to noise ratio)