Edge-Preserving Image Denoising via Optimal Color Space Projection

Denoising of color images can be done on each color component independently. Recent work has shown that exploiting strong correlation between high-frequency content of different color components can improve the denoising performance. We show that for typical color images high correlation also means similarity, and propose to capture inter-color dependency using an optimal luminance/color-difference space projection. Experimental results confirm that performing denoising on the the projected color components yields superior to existing solutions denoising performance, both in PSNR and visual quality sense. We also develop a novel approach to estimate directly from the noisy image data the image and noise statistics, which are required to determine the optimal projection. Index Terms – Image denoising, wavelet thresholding, luminance, color differences, optimal color projectio