Patches Based Multichannel Weighted Nuclear Norm Prior for Color Image Denoising

Image denoising is a fundamental problem in the field of signal processing and computer vision. A lot of work has been proposed for gray image denoising. Because there are different noise variances in different color channels, it is not easy to extend these methods for color image denoising directly. In this paper, we propose a novel method for color image denoising by introducing a multichannel weighted nuclear norm prior into the patches based image recovery framework. The different noise statistics of three color channels are taken into account by using a weight matrix to data fidelity in multichannel weighted nuclear norm prior. After each patch is updated, then the whole image can be recovered by global convex optimization. Each alternative updating step has a closed form solution and the problem can be solved efficiently. We demonstrate the superiority of proposed method over some state-of-the-art denoising algorithms on synthesis and real color image datasets.