A structure-preserved nonlocal iterative regularization model for image denoising

Non-local Means(NLM) is increasingly popular in image denoising. In this paper, the nonlocal structure similarity of images obtained by the iteration is exploited. By combining the nonlocal similarity constraints with total variation regularization, an iterative regularized variational model is proposed, in which the nonlocal weight depends on local structure of patches. An effective algorithm is also presented to solve the optimization problem based on split Bregman iteration. Experimental results reveal that the proposed method is competitive with the state-of-art denoising algorithms, especially for images with strong noise.