A noise estimation method for multiplicative noise removal

In this paper, a new splitting algorithms (i.e., noise estimation methods (NEM)) for multiplicative noise removal is proposed. By introducing a noise constraint with auxiliary variables, which guarantees the subproblem of the corresponding noise variable has a closed form solution. Further, we extended NEM to have a fully split form (NEMF). Specifically, add a proximal term to the subproblem of the original variable, which can improve the computational efficiency. Numerical experiments have proved that the proposed methods outperforms some state-of-the-art models in terms of the SNR.