A novel adaptive diamond search for fast block-matching motion estimation

Block-matching algorithm (BMA) is always the focus of video compression techniques in the past decades. In this paper, the focus is not played on the search pattern but on the search strategies. The proposed search algorithm has four novel points: new adaptive motion vector prediction, new zero MV priority, new adaptive search early termination threshold, and new adaptive search pattern based on sequences´ motion characteristics. We care all the performances including PSNR, number of search point (NSP) and bit-rate. Simulation results using extensive standard sequences have shown that the proposed search algorithm is superior to conventional fast BMAs because of its robustness, adaptability and efficiency.