A Novel Image Restoration Algorithm using Sliding Window and Neural Network

Abstract

As images become one of the most significant information carriers in daily life, it is essential to transmit and store high quality images. In the case that original image is degraded, a method to extract information and restore the degraded image is needed. In this paper, a novel algorithm is proposed to restore degraded image by using sliding window and neural network. In our algorithm, first, we used the sliding window to sample the image, then put the sampled sequence into neural network for training, establishing a nonlinear mapping model between the target clear image and the blurred image. Finally, we put the degraded image into the well-trained network to restore. The simulation results and analysis of the restoration on MATLAB are also showed, it turns out the quality and clarity of the degraded image are well improved.