A novel algorithm for color image denoising based on the CB color model

In this paper, we propose a novel algorithm for color image denoising. The algorithm is based on Chromaticity-Brightness (CB) color model which separating the color image into two components: chromaticity and brightness. The chromaticity is a unit vector on S². Then the vectorial ROF model with unit norm constraint is used to process the chromaticity component. We first use Lagrange multipliers method to deal with the unit norm constraint, and derive the closed-form solution of Lagrange multipliers. Secondly, we add a new variable to substitute the chromaticity such that the original problem is split into two easier subproblems. One subproblem is minimizing the standard vectorial Rudin-Osher-Fatemi (ROF) model which can be solved by generalizing the Shen´s fast operator splitting based scheme and the other has closed form solution. A number of experiments demonstrate that the proposed algorithm is effective in chromaticity and color image denoising.