一种双模式的运动估计算法

Double mode search algorithm for motion estimation

为了平衡运动估计中搜索算法的复杂度与搜索精度,基于双模式算法的思想,提出一种结合基于改进的粒子群算法(PSO)和十字搜索算法(ARPS)的双模式运动搜索算法。该算法对不同运动程度的图像采用不同的运动搜索算法(运动剧烈时采用PSO算法,运动平缓时采用ARPS算法),有效地结合了PSO的全局性特点以及ARPS的局部性特点,同时保持了ARPS的快速性。实验表明,该算法的整体性能高于传统的单模式运动估计算法以及已有的多模式运动估计算法。

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.