An adaptive motion search algorithm based on temporal-spatial correlation

This paper presents a new adaptive diamond search algorithm for motion estimation based on temporal-spatial correlation property. Each frame image taken from standard video sequences is separated to blocks, the macro-blocks are classified by a criterion based on temporal-spatial correlation value and the initial search point is then predicted. Different search modes are selected according to different image contents, and the best matching block is gotten in accordance with MAD rule. Experiment results show significantly improved performance in image quality and much reduction in computational complexity.