Spatially-adaptive wavelet denoising scheme with linear color shrinkage

To remove signal-dependent noise of a digital color camera, this paper presents a new color-image denoising scheme in a wavelet transform domain, which is newly developed from the classic spatially-adaptive wavelet denoising scheme with the Wiener-type linear shrinkage function. The classic Wiener-type linear shrinkage function works well for monochrome images; to utilize linear inter-channel color cross-correlations, usually a noisy input color image undergoes the color-transformation from the RGB color space to the luminance-and-chrominance color space, and then the luminance and the chrominance components are separately denoised. However, this color-denoising approach cannot take advantage of nonlinear inter-channel color cross-correlations caused by color edges, and cannot cope with actual signal-dependent noise. To utilize the noise's signal-dependencies, this paper constructs a new Wiener-type linear color-shrinkage denoising scheme where the inter-channel color cross-correlations are directly utilized in the RGB color space, and its shrinkage parameters are spatially-adaptively controlled according to not only local statistics in a noisy input color image, but also the noise's signal-dependencies. Experimental simulations demonstrate that our denoising scheme alleviates artifacts caused by denoising, and improves picture quality of denoised images, as well as we have expected.