Restoration quality assessment

It has been shown that restoration algorithms often compute error estimates which can be interpreted as restoration quality measures. These measures constitute useful information, which can be used during a verification phase or by other restoration algorithms. Two examples have been further developed, taking into account these ideas: a global motion estimation algorithm, and a local random defects restoration algorithm. They can be combined: the global motion estimation algorithm allows one to compensate for camera motion in an efficient way, and the blotch restoration algorithm not only detects blotches, it also applies a more accurate motion estimation procedure around each defect candidate in order to interpolate missing information while trying to detect possible false detections caused by small fast moving objects. (6 pages)