A novel fast motion estimation algorithm based on block-matching
Kexiang Liu, Qingshun Qiu, Zhenzhong Zhang · 2011
In order to reduce the computational complexity of motion estimation, a new fast and effective algorithm is proposed. This new algorithm is based on the real world image sequence's characteristics of center-biased motion vector distribution. It uses a small diamond search window to replace the small square search window used in the four-step search (FSS) algorithm. Simulation results show that this new algorithm provides competitive performance at faster speed when compared to similar techniques such as four-step search (FSS) and diamond search (DS).