Research On The Image Restoration Algorithms Based On Markov Random Field

There are a lot of factors such as the phase difference of the optical system, the atmosphere turbulence, moving diffusion of the focus and the system noise that degrade the digital images during their obtaining. The destination of image restoration is to recover image that has been degraded and make sure that the processed image as near as possible to the original image. In filtering of image, both denoising and maintaining of the detail about the image are required. But in most cases denoising of image conflicts with maintaining of detail in restoration processing. For the reason that mentioned above restoration of image is very challenge as a issue of digital image processing full of importance.At first, a model of image degradation and restoration is presented. In the following, some basic classic noise-reduction spatial filters are discussed briefly. And we proposed an adaptive filter used in situation of blended noise. The proposed method first classsifies the pixels into two classes, one is the pixels which are corrupted by Gauss noise and the other is the pixels corrupted by impulse noise , then applies the adaptive median filter and the adaptive weighted mean filter to removing impulse noise and Gaussian noise, respectively. Experiments show that our method is superior to the basic classic filters in image restoration. Next, the theory of Markov random field is introduced. Selecting a proper neighborhood system and using the ability of Markov random field to describe spatial dependence, MRF can be used to model the structural and characteristic of images. The restoration algorithm based on MRF look the original image as a MRF, and this is the prior knowledge of the estimation of Maximum a Posteriori (MAP). Which makes the difference is that how to calculate the MAP. Two traditional methods are used. One is a stochastic relaxation method, which is good at global optimality. The other is deterministic relaxation method, which is good at calculating speed. The paper provides a new changed simulated annealing method. The new method changes the J parameter dynamically during the iteration and divide the grid image into four plots. The experience results proof that the new method can restore the dirty images, which is added with Gauss noise, with higher speed than SA algorithm. Markov random field models provide a robust and unified framework for early vision problems such as image restoration. Inference algorithms based on belief propagation have been found to yield accurate results, but despite recent advances are often too slow for practical use. In this paper we introduce some algorithmic techniques that substantially improve the running time of the loopy belief propagation approach.