Therefore, in order to digitally restore a detected torn frame, the following tasks must be performed: 1) tear delineation, when detected torn frames are segmented into two regions divided by the tear boundary; 2) displacement correction, when the relative displacement between the segmented regions is corrected; 3) recovery of the damaged image data. The algorithm for the segmentation of torn frames is derived from interactive graph-cut methods used for still-image segmentation. We introduce new energy functions to describe both the region and boundary properties of the desired segmentation. The tear-delineation algorithm has a degree of flexibility built into it that allows user interaction to correct delineated frames when automatic processing fails.After tear delineation, the relative displacement between the regions is estimated using a modified global-motion estimation technique, which estimates the global motion of each region with respect to an unimpaired reference frame. Finally the degraded region is enhanced by means of patch filtering.