New total variation regularized L1 model for image restoration

In this paper, we propose three new time dependent models based on total variation regularized L1 model for solving total variation minimization problems in image restoration. The main idea is to apply a priori smoothness on the solution image and to regularize the parabolic term. We propose a kind of continuous boundary conditions, i.e. mean boundary conditions. Numerical experimental results demonstrate that the three new models need far less iterations than the original total variation L1 model and mean boundary conditions are efficient for image restoration.