Blind image restoration utilizing total variation regularization, shock filter and gradient reliability map

Tomio Goto, Hiroki Senshiki, Satoshi Motohashi, Haifeng Chen, Reo Aoki

Abstract: Blind image restoration, which restores a clear latent image from a single blurred image, is an ill-posed problem to find two unknowns: 1) the latent image, and 2) the point spread function (PSF). In this paper, we propose a novel PSF estimation method utilizing total variation regularization, a shock filter and the gradient degree of reliability. The experimental results show that our proposed method delivers the best performance out of several currently available methods with respect to restoration performance. Furthermore, the processing time for our proposed method is very quick, and clear images are obtained for actual pictures.