基于遗传搜索和模板匹配的快速运动估计

Fast Motion Estimation Based on Genetic Search and Pattern Matching

基于遗传算法的运动估计具有较好的全局寻优能力,但其过高的算法复杂度需要很大的计算和存储开销,增加了编码时间;另一方面,传统的基于遗传算法的运动估计普遍采用较低的遗传迭代次数,降低了遗传算法的搜索精度。为解决传统算法搜索时间长和搜索精度低的缺陷,提出了一种基于遗传搜索和模板匹配的混合算法。该算法结合多种运动矢量的预测方法以及运动矢量的统计特性设计了3种终止策略,使用模板匹配算法对遗传搜索过程进行优化。实验结果表明:该算法在确保编码质量的同时,能大幅缩减搜索点数和编码时间。

The motion estimation based on genetic algorithm has better global optimization ability,but its higher complexity enhances the cost for computation and storage and further increases the encoding time.On the other hand,the traditional motion estimation based on genetic algorithm usually adopts lesser genetic iterations,which reduce the searching precision.In order to solve the defects of longer search time and lower accuracy in traditional algorithms,this paper proposed a hybrid algorithm based on genetic search and pattern matching.According to the statistical properties of the motion vector and existing motion vector predicting methods,three termination strategies were designed in the present algorithm,and the matching algorithm was also adopted to optimize the genetic search process.Experiment results show that the proposed algorithm can greatly reduce the search points and coding time while attaining good quality in coding process.