Image Diffusion Model Algorithm, The Split Bregman,

Image diffusion is one of base research field in image processing. The majority application field is in image cover, image painting, image denoising and so on. Image diffusion based on variational methods and PDEs is a basic methods of image restoration. The basic idea of variational methods is take the question as calculation of the minimum of a functional. And then, we get the numerical solution through solving the PDEs which are derived from the energy functional. In this paper, firstly we introduce the module of the image diffusion simplify, and then introduce Charbonnier model, generalized TV model, P-M model as examples. Some numerical examples are given to show their capabilities of forward/backward diffusion and edge preserving enhancement. At last we propose a new method that is Split Bregman to use in image diffusion model. Firstly, we used this method in TV models for solving minimization problems. Then it extended to the general nonlinear model. By introducing the shrink operator for reduce the computational, and because we used Bregman distance in our method that could improve the image. Finally, we used experiment to verify its feasibility and effectiveness.