Adaptive Multiplicative Noise Removal Model Using Pixel Related Smoothing Term

Abstract

Multiplicative noise removal problems have attracted much attention in recent years. In this paper, we propose a new adaptive multiplicative noise removal algorithm based on variational method. By analysis the shortcoming of Euler- Lagrange equation, we find that these traditional variational models are not fitted for multiplicative noise very well. The amount of multiplicative noise is relative with the pixel value. That is to say, areas with large pixel value should be smoothed much than areas with small pixel value. So we modified the Euler-Lagrange equation by changing the balance parameter to the pixel related one, then we deduce our modified energy equation. The balance of fidelity term and regular- ization term in our changed model can be changed in different areas with different gray value. Thus in iterating procedure, our method can change the degree of noise removing in different areas with different noise level adaptively. It can also preserve the edges and remove the noise very well. The results show the outperforming effect of our method.