Chrominance restoration from luminance by statistics matching

This paper deals with a problem existing in art restoration. More precisely the algorithm that is being presented tries to restore chrominance information onto a gray level image as accurately as possible. The method was developed to try to restore the information of baroque paintings on the ceiling of a church. Those paintings were severely damaged during the Spanish Civil War and the only available information is: (i) an (geometrically distorted) ancient gray level photograph of the whole original painting. (ii) Today's partial pictures of the remaining portions. (iii) Intact paintings of the same artist, period, and style in other churches. The objective is then to try to obtain as good a hypothesis as possible of the original artwork, using all the available information.