Image Inpainting By Robust Principle Component Analysis(RPCA)

Image inpainting is the process of rebuilding lacking or damaged parts of a picture depended on prior knowledge or background information.The traditional methods just use the Local information of a image to recovery,so the results of recovery was always not well.Nowadays,lots of people starts to use the algorithm of Robust Principal Component Analysis(RPCA)for image inpainting,but when traditional rank minimization is applied to these problems,the expression of the objective function is not make the best of the previous objective rank in-formation on these problems.In the thesis,I use a low-rank structure with global information to fix the image,the Augmented Lagrange Multiplier(ALM)to formulate my algorithm,and optimization with Alternating Direction Method of Multipliers(ADMM).With analyze my experimental results,I conclude that the information of picture scenes is controlled by the largest singular value of the previous small part.Inspired by this,I use the characterize top 20 largest singular values of the information to repair the image.My experimental analyses show that,using the method to repair images is better than previous traditional method.