基于运动信息自适应的运动估计算法  

Motion Estimation Algorithm Based on Motion Information Adaptation

作　　者：林芳明[1] 

机构地区：[1]江西理工大学信息工程学院,江西赣州341000

出　　处：《科技广场》2015年第3期29-33,共5页Science Mosaic

基　　金：江西省研究生创新基金(编号:YC2013-S199)

摘　　要：结合分析UMHexagon S和经典运动估计算法,本文提出了一种基于运动信息自适应的运动估计算法。该算法利用宏块的时空相关性实现对静止块的判定,并对其直接停止搜索;依据块的运动类型来自适应选择起始点和模板,对于中、小运动块跳过大模板粗搜索直接进入小模板细搜索。实验结果表明,基于运动信息自适应的运动估计算法的搜索精度接近于UMHexagon S,但是搜索速率优于H.264标准中已有的快速运动估计算法。Based on a comprehensive analysis of UMHexagonS algorithm and classical block matching mo-tion estimation algorithm, the paper proposes a motion estimation algorithm based on motion information adapta-tion. Based on the space-time relativity of macro block, the algorithm achieves static block determination and can end the search static block directly; According to the type of movement, the algorithm can adaptively select initial searching point and templates. As for small and middle sized motion blocks, it will skip a big template coarse search-ing and go directly into the small template fine searching. The experimental results show that the searching accuracy of the motion estimation algorithm based on motion information adaptation is close to UMHexagonS algorithm, but its searching speed is better than that of the existing fast motion estimation algorithm in H.264 standard.

关 键 词：运动估计 时空相关性 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]