A new anisotropic diffusion model for image denoising

In this paper, we propose a new algorithm for image denoising. Our method is an anisotropic diffusion model, and it is based on variation and partial differential equations. The new model not only overcomes the shortcoming of TV model, which does not fully comply with the morphologic principles of image processing, but also overcomes the shortcoming of ALM model, which depends on gradient information alone to determine the speed and direction of diffusion. Numerical results show that our model can maintain the main features of the image. The denoising effect of our model is better than some traditional models. The main feature of our new model is making the processed image retain edge feature and corner information of the image.