Study On Image Denoising Based On Variation And Partial Differential Equations

In recent years, the image processing based on variation and partial differential equations has achieved rapid development and has become an important method in the field of image processing. Image denoising is a very important part of the image processing, which is of great significance for improving the visual effect and quality of the image. In this study, we mainly research the image denoising models based on variation and partial differential equations and analyze these models. On the basis of previous knowledge, we put forward the improved models. The main contents is organized as follows:The main innovation and contribution of this paper is that we obtain new denoising models by combining different models. We analyze TV denoising model,1TV-L denoising model and OSV denoising model respectively. TV denoising model is a classical model of image denoising, which has opened up a new way for image analysis,but it exists some defects at the same time.1TV-L model not only can reduce staircase effect and the loss of geometric feature, which are obtained by TV denoising model, but also can improve the numerical stability. However, it is difficult to solve1TV-L model.Because of the simplicity and effectiveness of the split Bregman iteration algorithm, it can be used to solve1TV-L model. In addition, OSV model can decompose a picture contains texture into the cartoon part and the texture part, result in more accurate image edge information form the cartoon section without influence from the texture section, so that it has good performance of image restoration. However, it is inefficient to solve OSV model by traditional algorithm, we can also use the split Bregman iteration algorithm to solve this problem. On the basis of previous analysis, the OSV denoising model is combined with the anisotropic denoising model and isotropic denoising model respectively. By changing2 L norm of the residual item to1 L norm, we get two improved denoising models. Similarly, the split Bregman iteration algorithm is applied to solve these two improved models. Though the subjective and objective evaluation method and numerical experiments of MATLAB 7.0, we can see that the denoising effect of the improved models is better than that of the original models.