A new full-reference measure of color/detail preservation is presented in this paper for the validation of noise reduction techniques in color image processing. The method, namely Color Peak Signal-to-Blur Ratio (CPSBR), extends to the RGB color space our previous approach to the estimation of filtering blur in grayscale pictures. The accuracy of the proposed approach is investigated considering an important and widespread used nonlinear technique for noise removal: the bilateral filter. The exact values of color distortion and detail blur produced by this filter are theoretically evaluated in the paper and used for a comparison. Computer simulations dealing with color images corrupted by different amounts of Gaussian noise and processed by bilateral operators with different parameter settings show that the proposed method yields very satisfactory results.
