基于预测搜索距的快速块匹配运动估计算法

Search-length based fast block-matching motion estimation algorithm

提出了一种基于搜索距的快速运动搜索算法。该算法采用多个初始预测候选运动矢量集进行运动估计,通过对与该宏块时空域相邻宏块的搜索距来预测当前宏块的搜索距,根据当前宏块搜索距和块失真特性统计的不同动态采用不同的搜索模板。模拟试验表明:提出算法能获得更好的视频质量,拥有良好的搜索速度伸缩性,优于传统的PMVFAST算法。

A search-length based on the fast block-matching motion estimation algorithm is proposed. The proposed algorithm uses multiple initial predictor candidates and the predictor SL and SAD of neighbor macroblocks in space and temporal correlation, according to the different statistical SL and SAD, the proposed algorithm uses different search criteria and search pattern. Extensive simulation results demonstrate that the proposed algorithm is better than other recently proposed algorithms and it can control the search speed according to practical demands.