A Fast Algorithm for TV-L1 Based on Split Bregman Method

摘要

The Rudin-Osher-Fatemi (ROF) model is one of the best successful models for image denoising. Recently, the TV-L| model, which replaces the L2norm of the standard ROF model with the L1-norm, has been proposed. The L1-norm outperforms the L2-norm for some applications. However, because of the total variation (TV) norm and Li-norm, it is difficult to compute. In this paper, we propose a fast algorithm based on the Bregman / Split Bregman method to solve the TV-L1 minimization problem, and show several examples to demonstrate its efficiencies.