Study On Restoration Of Cultural Relics Image With Rank Minimization

With the rapid development of information technology,digital image processing technology has been everywhere and widely used in every field of human society.Cultural relics images are affected by all kinds of natural factors for their long history,which makes cultural relic face noises,scratches,loss of image patches and so forth that cause cultural relics cannot be presented completely;Repairing them manually needs experts and has such drawback of long period.Therefore,repairing cultural relics with image restoration technology has deep impact.Starting from the prior model of Rank Minimization,a research in depth concentrated on the problems of cultural relics image denosing and restoration is executed.The main achievements of this paper are as follows:Firstly,This article gives an overview of existing image restoration methods.In terms of image denoising,the nonlocal average,K-SVD and BM3 D denoising algorithms were described,and the principle of each algorithm was introduced in detail;In terms of image restoration,the BSCB,TV and CDD classic repair methods were mainly introduced,the formula of each algorithm was deduced and the principles were analyzed.Secondly,on the cultural relics image denoising,a cultural image denoising method based on the non-local RPCA was put forward.This method used nonlocal idea to find similar blocks and form a similar block group matrix.For the similarity among similar blocks,if Robust Principal Component Analysis was used directly on it to low-rank denose,the similarity of two adjacent blocks was not taken into account.For this consideration,a penalty term denoting the similarity of two adjacent blocks was add to initial RPCA model to conduct better denosing on cultural relic images.In the process of solving the formula,in order not to increase auxiliary variable and the amount of computation,linear alternative direction was adopted to get optimal solution.The experimental results show that,compared with the contrast method,this denoising method proposed here has better ability to keep information and achieve better denoising effect.Thirdly,On the cultural relics image restoration,this paper proposes a restoration method based on the non-local matrix completion.The method use the ideas of non-local to join matrix completion to the restoration algorithm;makes full use of the redundancy of the image information to look for a similar blocks that constitute a similar set of matrix in the search window and uses the principle of matrix completion to repair the image.Expriments on cultural relics with combination of other state-of-the-art restoration methods show that this algorithm can maintain the basic information of the image and structural characteristics maximally and get a higher PSNR and SSIM values;to a certain extent,can well guarantee high visual quality of the restored image.