Abstract:
Center-biased characteristics of motion vector probabilities distribution is very important for motion estimation, but directional feature of motion vector probabilities distribution is ignored. An unsymmetrical cross-search algorithm (UDCS) is proposed. Search steps of using small cross-search pattern to find small motion vectors in initial phase and unsymmetrical double cross-search pattern to further determine search large motion vectors in subsequent steps are brought forward. Finally, architecture of the algorithm for implementation is introduced and performance is analyzed.
Key words:
Block-matching motion estimation,
Unsymmetrical cross searching,
Cross-center-biased property
摘要: 视频序列运动矢量分布的中心偏移特性在运动估计中发挥重要的作用,但忽视了运动矢量分布的方向性。该文提出了非对称双十字形搜索(UDCS)算法,给出了在搜索的初始阶段使用小十字搜索模型对小的运动矢量搜索并在相继的搜索过程中使用具有方向性的十字形搜索模型对大运动矢量进行搜索的步骤。介绍了该算法的实现结构,并分析了该算法搜索性能。
关键词:
块匹配运动估计,
非对称十字形搜索,
十字中心偏置特性
CLC Number:
LIU Haihua;CHEN Xinhao;GAO Zhiyong. Study and Implementation on an Unsymmetrical Double Cross-search Algorithm for Block-matching Motion Estimation[J]. Computer Engineering, 2006, 32(15): 200-202,.
刘海华;陈心浩;高智勇. 非对称双十字形运动估计搜索算法研究与实现[J]. 计算机工程, 2006, 32(15): 200-202,.