Adaptive motion tracking based on spatial correlation

An adaptive motion tracking scheme based on spatial correlation is presented. By the property of the magnitudes of motion vectors (MVs) of neighboring blocks, the current MV is first estimated as the initial search centre which may be close to the global minimum, and then refined by certain fast search algorithm. In order to avoid error propagation, those frames whose distortions exceed tolerance are to be searched afresh for MVs using a fast search algorithm. Experimental results show the proposed scheme improves the performance of estimation with lower computational complexity.