METHOD AND APPARATUS FOR DETECTING OPTIMUM MOTION VECTORS BASED ON A HIERARCHICAL MOTION ESTIMATION APPROACH

摘要

Motion vectors are determined between a current frame and a preceding frame within a frame duration by employing a multi-tier hierarchical motion estimation approach, wherein the candidate blocks are grouped into a multiple number of hierarchical search areas. Candidate blocks contained in a search area of a highest hierarchy are examined first to check if a minimum means square error (MSE) between the candidate blocks and the search block is smaller than a predetermined threshold. If the minimum MSE meets the requirement, the position vector of the corresponding candidate block is determined as the motion vector of the search block; or else, candidate blocks contained in a next hierarchy are processed. To find a motion vector of the search block, different requirements are imposed on the search areas of remaining hierarchies. One of the requirements is that an MSE value for a motion vector should be smaller than a reference error function which is defined as a minimum among MSE values of predetermined candidate blocks. The other requirement is that the minimum MSE value of a currently processed hierarchy is much smaller than the minimum MSE of higher hierarchies; or a minimum MSE value increases twice consecutively between 3 adjacent hierarchies. The procedure continues until a motion vector which meets the requirements is found or all hierarchies are processed.