The Image Restoration Method Based On MCP Function For Impulse Noise

Digital image processing can not only improve the image quality to meet the requirements of people on the vision,and has important engineering applications in many scientific and engineering fields,such as weather forecast,industrial production,remote sensing technology,biological medicine and communication.In this paper,we study the image restoration problem with fuzzy polluted and high impulse noise.Although there are a number of models to remove impulse noise,especially models based on the Rudin-Osher-Fatemi(ROF)model,their excellent edge-preserving property is incomparable to the other models.In the existing methods,Fast Total Variation Deconvolution(FTVd)method,Two-Phase and Total Variation-l0(TVL0)model are effective ways to remove impulse noise while the random noise level not higher than 50%or the salt-and-pepper noise level not higher than 70%.In this paper,we propose a new method based on Minimax Concave Penalty(MCP)function,so we call the model TV-MCP.Alternating Direction Method of Multipliers(ADMM)method is used to solve this model,and the global convergence has been proved.In numerical experiment,we apply FTVd,Two-Phase,TVLO and TV-MCP to the problem of image de-noising and de-blurring in the presence of impulse noise.The results show TV-MCP outperforms the other three methods especially for the high noise level image de-noising.The author's major contributions are outlined as follows:1.From the perspective of' high level removal of impulse noise,based on the total variational model,the data fitting with MCP penalty function,this paper proposed a new impulse noise image restoration model,called the TV-MCP model.2.The DC proximal method of TV-MCP model is given,also the ADMM of the sub-problem.The numerical results show that compared with the existing method to impulse noise with fuzzy,TV-MCP model is not only has advantages on high impulse noise level with fuzzy image restoration problem,for less than 70%of the salt and pepper noise and random values below 50%noise image restoration are also superior to other methods,are very effective in ascension.3.Proved the convergence of TV-MCP model and the DC proximal method.