Comparison of image restoration and segmentation of the image using neural network

Abstract

In present day almost all of the image restoration method suffer from weak convergence properties. Also for Point Spread Function (PSF), some methods make restrictive assumptions. Some original images restrict algorithms portability to many applications. Current situation is using deburring filters, images are restored without the information of blur and its value. In this paper, method of artificial intelligence is implemented for restoration problem in which images are degraded by a blur function and corrupted by random noise. This methodology uses back propagation network with gradient decent rule which consists of three layers and uses highly nonlinear back propagation neuron for image restoration to get a high quality of restored image and attains fast neural computation, less complexity due to the less number of neurons used and quick convergence without lengthy training algorithm. The basic performance of the neural network based restoration along with segmentation of the image is carried out.