Image denoising considering characteristics of YCbCr color channels

In this paper, we propose a channel-adaptive denoising framework that considers YCbCr color space. The framework uses non-local means filtering for the Y channel and anisotropic diffusion filtering for the CbCr channels. Compared to non-local means filtering, which gives the highest peak signal-to-noise ratio (PSNR) among weighted average methods, the proposed approach reduced the processing time by ~ 76 % and reduced energy consumption by ~79 %, while maintaining comparable PSNR. The proposed approach is expected to be used in low-power, high-speed denoising frameworks.