Convex and non-convex adaptive TV regularizations for color image restoration

Color image restoration is an important and challenging research topic in image processing. Different from grayscale images, each color image has three channels in RGB color space. Due to the correlation among the three channels, color total variation (TV) regularized image restoration based on the local channel coupling is better than the direct application of its grayscale counterpart in each channel of color images. On the other hand, an adaptive weighting scheme is a good technique for restoring local features of images. Inspired by these two strategies, we propose convex and non-convex adaptive TV regularized models for color image restoration to better handle image local features. Numerically, we design an alternating direction method of multipliers to efficiently solve the proposed two models. Comprehensive experiments are conducted to demonstrate the effectiveness and advantages of the proposed methods.