Image Restoration And Enhancement Based On Markov Random Field

Image information, with the advantages that can carry large amounts of information, transmit in long distance and can be easily analysed directly, is more and more widely used in areas like our daily lives, national military and industry However, in practical application, the directly obtained image information is often affected by the interference of distortion like noise, blur and so on, because of limitations of shooting equipments, weather factors and harsh natural conditions. In this situation, the further analysis of image information will be affected, that’s why image processing of directly obtained image is a research topic of great significance.Technically, the clarity processing of directly obtained image mainly has two aspects:image restoration and image enhancement. Nowadays, with many years’ development, there has been lots of mature traditional technology for image restoration and image enhancement including filter recovery, wavelet enhancement and so forth. The theory of Markov random field (MRF) is raised after the Markov property is generally studied.After more and more researches and applications in recent years,MRF have been widely used in image segmentation and denoising direction,but still not so mature in image enhancement.So this paper is aiming at using the MRF in image restoration and image enhancement,thus to improve the method.In this paper, firstly, the basic theory of MRF is described, and several commonly used MRF models are introduced as well as the ways to solve these models. Then, two novel MRF models: DAMRF model and QMRF model are introduced. In this paper, based on the theoretical knowledge, the MRF is used in image restoration and an improved random relaxation algorithm is raised. Also, the algorithm is compared with the traditional algorithm and simulated annealing algorithm separately. The results show that this algorithm has better visual effect compared with the traditional algorithm and better convergence compared with the simulated annealing algorithm. Then, the paper presents a image enhancement algorithm based on the DAMRF model, which uses MRF in image enhancement. Through experiments, we can see this algorithm has a better edge enhancement effect than the traditional algorithm. Finally, we summarize the algorithms for image restoration and image enhancement based on MRF, and look ahead the future prospects of this method’s development.