Image Denoising Based On Nonlocal Partial Differential Equation

In the process of image capturing and transmission, noise is always introduced,which has great influence on the processing and storage of image information.Therefore, a large number of researchers devote themselves to research to find a method of removing noise effectively and keeping details simultaneous. Denoising algorithm based on nonlocal Partial Differential Equation is obtained by combing nonlocal means denoising algorithm and PDE based on image denoising. This method is a powerful tool to remove noise. With the development of denoising algorithm based on nonlocal Partial Differential Equation arises at the historic moment. Denoising algorithm based on nonlocal Partial Differential Equation is widely used to image processing. Compared with the traditional denoising method, this method has advantages in image denoising.It can not only eliminate noise but also avoid staircase effect and speckle effect.Based on the profound analysis on nonlocal PDE denoising algorithm, several classical PDE based on image denoising and nonlocal means denoising algorithm are introduced in detail, and this paper introduces several common nonlocal PDE denoising algorithm in details. Finally, I compared with these methods, as well as discuss their advantages and disadvantages respectively, results and conclusions are given at last.In these nonlocal PDE denoising algorithm, this paper gives two models. One model is image restoration combining the PM model and the LLT model, another model is image denoising combining the special PM model and the nonlocal denoising algorithm. The major content of this paper includes some following aspects:(1) Firstly, I have given a solution of second-order PDE denoising model and a solution of fourth-order PDE denoising model, then combine them to an effective model.This new model can not only avoid staircase effect, but also remain the image details.(2) First of all, an unusual model is obtained by changing the diffusion coefficient of P-M model, then combined with nonlocal denoising model and applied to the color image. For the color image consisted of three channels with red(R), green(G) and blue(B), in the process of denoising, we can restore the image with better quality by starting with the gray scale image and using the fast Bregman Split iterative algorithm to remove the noise. Finally, the denoising effect of the model is verified by experiments.