New time dependent pretreat models based on total variational image restoration

Abstract

In this paper, we propose new pretreat models for total variation (TV) minimization problems in image deblurring and denoising. Specially, blur operator is considered as useful information in restoration. New models in form is equivalent to pretreat the initial value by image blur operator. We successfully get a new (L. Rudin, S. Osher, and E. Fatemi) ROF model, a new level set motion model and a new anisotropic diffusion model respectively. Numerical experiments demonstrate that, under the same stopping rule, the proposed methods significantly accelerate the convergence of the mothed, save computation time and get the same restored effect.