Colorimetric restoration of digital images

A colorimetric approach to restoration of digital images is presented. Assumptions are made to simplify the general problem to obtain a more computable form. Two methods are developed, using Karhunen-Loeve transformation and independent restoration schemes from earlier works, to solve the estimation problem in color image processing using multidimensional restoration. A comparison of the methods is presented including the effects of parameters of interest to desktop scanners and digital cameras. The results for the SNRs and blurs studied indicate that more than three color channels produces a slight numerical gain and modest visual gain.