A variant beltrami flow for multiplicative noise removal

In this paper, we propose a new diffusion approach for multiplicative noise removal. The diffusion is driven by two terms. One is the regularization term which comes from the Beltrami flow, the other is the fidelity term inspired by the Aubert-Aujol (AA) model. The two terms are balanced by a weight parameter. In order to overcome the difficulty in choosing the best weight, we derive an automatic scheme. Numerical results show that the proposed method preserves edges better than the scalar AA model while smoothing out the multiplicative noise.