Numerical scheme for efficient colour image denoising

In this paper, we are interested by the enhancement of colour images, where we present a numerical scheme to implement a non-linear diffusion filter. This scheme is developed to denoise colour images corrupted by additive noise. The method is based on harmonic averaging that takes into account correlation between all colour components of the image by using a common gradient magnitude which can be computed by using the Di Zenzo idea, and a marginal gradient magnitude in order to conserve the self behaviour of each colour component. The proposed scheme is an efficient tool at image selective smoothing in presence of strong noise. This analysis shows that our method performs better than some related works, and particularly in avoiding colour artifacts.