Algorithm For Image Restoration Based On Multi-Grid Method

Recently partial differential equations (PDE) are used to restore noising images and obtain good results. In this paper, we first describe some image restoration methods based on PDE models and the problems caused by the models. And then we proposed two new improved methods to figure out the problems. At last, we use multi-grid numerical solutions to solve PDE Models. Our research focus on the following methods: 1) An improved method is proposed by adding the same direction gradient diffusion to the Perona-Malik model. The experimental results show that the isolated noise points in the original images can be removed more effectively with the edges keeping well. 2) A coupled PDE method for image denoising is proposed based on the Perona-Malik model and fourth order PDE. And the experimental results show that the new method can remove noises while preserving edges well and avoid blocky effects which is easily seen in images processed by second order PDE. 3) Multi-grid not only provides a significant acceleration, but also increases the calculative precision. So we proposed the same direction gradient diffusion based on multi-grid method. The similar algorithm is applied to the model of the coupled PDE method. Numerical experiments show that the new algorithm is efficient and more time-saving.