Adaptive motion estimation based on statistical sum of absolute difference

A novel motion estimation algorithm for video compression, AMESSAD (adaptive motion estimation based on statistical sum of absolute difference) is presented. The algorithm adaptively determines motion search window size based on short-term and long-term statistical distribution of motion activities. It performs motion estimation efficiently for hybrid video encoders, compared with full and fixed-step search BMA algorithms. In addition, improved perceptual quality can be obtained over fixed step search because the algorithm differentiates regions with motion content and provides better coding quality to local macroblocks and frames that have richer motion content