Virtual restoration of old mural paintings using patch matching technique

Changes occurring in nature cause considerable deterioration of Old mural Paintings. This paper presents a virtual restoration scheme to restore the degraded versions to some extent. Our scheme is based on the assumption that there exist mural Paintings which are completely in their original form, without any degradation. The scheme then presents a technique of best patch matching from a database obtained from clean paintings, where the error patch is replaced by the best patch found. The above is done taking into account the time of execution of the whole process. Prior to patch matching, image enhancement techniques are used on the image to ensure better patch matching. Finally post processing operations are performed on the output for the final cleaned mural. This paper includes demonstration of the above mentioned algorithm on real images, which shows the validity of the proposed scheme.