Improved motion-compensating prediction with rotational block matching

Motion compensation is one of the essential components of moving-video coding for the exploitation of statistical redundancy. Since the prediction in the moving areas of a scene can be improved by motion compensation, this technique is applied in almost all low-bit-rate image sequence coders. In the past, several algorithms have been developed for estimation and compensation of motion. Most of these approaches calculate the motion vectors either by evaluating spatio-temporal gradients in a differential approach, or comparing shifted image parts in a block matching approach. The aim of this contribution is to show the gain which can be obtained by additional compensation of rotational motion in comparison with the compensation of translational motion alone. A consideration of side information, which has to be transmitted, is also included