The Generalized MTV-L1 Models and Their Split Bregman Algorithms for Noise Removal of Color Images with Textures

The classic TV(Total Variation)-L1 model for gray value image denoising is extended to a generalized TV-L1 model firstly, which is then extended to generalized MTV-L1 model for color image denoising based on MTV(Multichannel Total Variation) regularization. By introducing the concept of non-local MTV based on non-local means, the generalized non-local MTV-L1 model is designed for noise removal of color images with textures. The Split Bregman algorithms are designed for the above mentioned models proposed in this paper. Finally, some numerical examples are provided to validate the proposed models and algorithms.