Double mode search algorithm for motion estimation
一种双模式的运动估计算法*

To balance the complexity and accuracy of motion estimation, based on the idea of double modes, this paper pro-posed a search algorithm which combined an improved particle swarm optimization (PSO) and adaptive rood pattern search (ARPS). The algorithm used different mode for different cases according to the level of motion (ARPS for mild motion and PSO for drastic motion). It combined the global feature of PSO and the individual feature of ARPS and kept the fast feature of ARPS. The simulation results show that its effectiveness is higher than that of single mode algorithm and other double mode algorithm which have been proposed.