Image Restoration Based on Bi-Regularization and Split Bregman Iterations

In this paper, an efficient approach for image restoration is proposed. Our method combine the regularization based on sparsity and split Bregman iteration techniques. We employ bi-regularization based on curvelet and DCT to constrain structure and texture components of restored image respectively. The experiments show that the proposed approach can well recover edges and most of the details of a textured image. Hence, bi-regularization and the split Bregman iteration are efficient for image recovery.