Markov Random Field based Image Restoration with aid of Local and Global Features

Abstract

restoration is the process of renovating a corrupted/noisy image for obtaining a clean original image. Numerous MRF based restoration methods were utilized for performing image restoration process. In such works, there is a lack of analysis in selecting the top similar local patches and Gaussian noise images. Hence, in this paper, a heuristic image restoration technique is proposed to obtain the noise free images. The proposed heuristic image restoration technique is composed of two steps: core processing and post processing. In core processing, the local and global features of each pixel values of the noisy image are extracted and restored the noise free pixel value by exploiting the extracted features and Markov Random Field (MRF). Moreover, the restored image quality and boundary edges are sharpened by the post processing function. The implementation result shows the effectiveness of proposed heuristic technique in restoring the noisy images. The performance of the image restoration technique is evaluated by comparing its result with the existing image restoration technique. The comparison result shows a high-quality restoration ratio for the noisy images than the existing restoration ratio, in terms of peak signal-to-noise ratio (PSNR).