摘要： 提出了一种基于六边形的运动矢量场自适应搜索算法,该算法充分利用了序列图像中运动矢量场的中心偏置性和时空相关性,结合六边形搜索法和小菱形搜索法,根据运动类型自适应的选择搜索策略和搜索起始点,同时设定阈值对静止块直接中止搜索。实验结果表明,该算法的搜索速度优于现有的快速运动估计算法,搜索精度接近于全搜索法。

关键词: 运动估计, 块匹配算法, 视频编码, HMVFAST

Abstract: This paper proposes a fast motion estimation algorithm, or the hexagon-based motion vector field adaptive search technique (HMVFAST). In HMVFAST, the type of local motion activity and initial search center are first determined according to the motion vector field, then a new hexagon search pattern and a common diamond search pattern are adaptively used. Experimental results show that the proposed algorithm provides faster searching speed than other existing fast block-matching algorithm, while the distortion is almost the same as the FS algorithm.

中图分类号: 

• TN919.81