Semi-blind image restoration based on Chan-Vese denoising model

A semi-blind image restoration algorithm is proposed based on reduced non-convex approximation of Luminita Vese and Tony Chan's (C-V) denoising model. Compared with C-V denoising model, we modify the fidelity term and add a term on point spread function (PSF). The function depends on two variables: the image function to be restored u and the standard deviation of Gaussian kernel to be estimated a. Then the problems consist in solving a system with two coupled equations. Compared with the Leah Bar's semi-blind image restoration model which must solve three coupled equations, our method only needs to solve two equations. Furthermore, the estimation of f by our algorithm is superior to Leah Bar's algorithm. The experimental results demonstrate that the proposed method is effective.