A novel parallel fast motion estimation algorithm
Wenying Yue, Nan Zhang, Baocai Yin · 2005
In this paper, a novel motion estimation algorithm is proposed. Candidate points are located in three regions, where search takes place parallel. The proposed algorithm can conquer the problem of local minimum effectively and reduce the computational complexity greatly, whose performance is comparable to that of the full search and the UMHexagonS algorithm used in H.264. The computer simulation results of the proposed technique compared with full search and other fast motion estimation algorithms are given for various sequences.