A Research Of Image Denoising And Restoration Based On Total Variation Method

The total variational method is a function which is through the introduction of energy ,making the image restoration problem transformed into a functional extremum problem. To study the functional extremum problem, this method is widely used in image processing. In this paper, the variational method based on the overall image denoising and image restoration of research.In this article, the total variational algorithm basic knowledge and the model of TV image restoration will be introduced firstly. In the part of the image denoising, the advantages and disadvantages of the isotropic diffusion model will be compared with TV denoising model and the adaptive TV denoising model. Through studying the varying regularization parameter, Response to the disadvantages of the above model, combined with the advantages of median filtering model to remove isolated noise, A new adaptive total variation median filter model which is based on gradient information are proposed. In the new model, noise is removed while the image edge information is preserved. In the same time, the generation of false edges is reduced and higher quality images are obtained.In the part of image defuzzification, first study the processing results of the TV model when the point spread function is known, and good experimental results are obtained. When the same parts of the point spread function are unknown or known, the restoration problem is named image blind restoration. When using the Blind Image Restoration, the TV model with a dual regularization parameter will be used, which guarantees the robustness of the image during the recovery process. First, according to the priori knowledge and the nature of the point spread function, an estimate value of the point spread function will be obtained. The image blind restoration problem is converted into a known blur kernel image restoration problem. As the recovery results were influenced by the ringing effect, the result is not very good, but it also significantly improves the image quality.