An efficient vector ranking filter for colour image restoration

An efficient ranked-order filter is proposed for removing the additive noise from colour images. In the approach presented here, vector ranking reduces to ordering vectors according to their distance to a centroid criterion, such as the sample mean. This technique significantly reduces the computational burden. Using this vector ranking scheme, α-trimmed mean filtering is considered. It can suppress both impulsive and Gaussian noise better than the median filter. A significant reduction in the complexity compared to the regular vector α-trimmed mean filtering is shown. A number of simulations were performed in order to quantitatively evaluate its performance. These simulations involve the estimation of colour images in additive short tailed and long tailed noise