Adaptively weighted difference model of anisotropic and isotropic total variation for image denoising

This paper proposes a novel nonconvex regularization functional by using an adaptively weighted difference model of anisotropic and isotropic total variation. By choosing the weights adap-tively at each pixel, our model can enhance the anisotropic diffusion so as to achieve robust image recovery. Regarding to numerical implementations, we express the proposed model into a saddle point problem with the help of a dual formulation of the total variation, followed by a primal dual method to find a model solution. Numerical experiments demonstrate that the proposed approach is superior over several gradient-based methods for image denoising in terms of both visual appearance and quantitative metrics of signal noise ratio (SNR) and structural similarity index measure (SSIM).