基于块特性与自适应搜索窗口的运动估计算法  

Motion Estimation Algorithm Based on Block Characteristic and Adaptive Search Window

作　　者：赵永利[1] 陈进成[1] 马健[1] 朱宝忠[1] 张杰[1] 

机构地区：[1]北京邮电大学光通信与光波技术教育部重点实验室,北京100876

出　　处：《数据采集与处理》2008年第3期273-277,共5页Journal of Data Acquisition and Processing

基　　金：国家高技术研究发展"863"计划(2006AA01Z244)资助项目;教育部新世纪优秀人才支持计划(NCET-05-0112)资助项目

摘　　要：提出了一种基于块特性与自适应搜索窗口的运动估计算法(Motion estimation algorithm based on blockcharacteristic and adaptive search window,MBC-ASW),该算法在充分利用视频图像的时间、空间相关性的同时,根据运动向量的统计特性,分别在帧层和块层进行自适应搜索窗口的粗调与微调,并且针对不同的块进行相应的编码处理。在PC上利用C语言实验验证了该算法的可行性,评估了其性能,并与经典的全搜索算法(Fullsearch motion estimation,FS)、三步搜索算法(Three step search,TSS)、新三步搜索算法(New three stepsearch,NTSS)和钻石搜索算法(Diamond search,DS)进行了详细的比较,该算法在性能和效率上均有较大程度的提高,对不同的序列具有较强的自适应性。A motion estimation algorithm based on block characteristic and adaptive search window （MBC-ASW） is proposed. While using the time and space pertinence of video, according to the statistic characteristics of motion vectors, the algorithm can adaptively adjust the search window on the layers of the frame and the block. Also different strategies are used for different blocks. The algorithm is realized and validated to be feasible on the computer based on C language, and the performance of the algorithm is evaluated. Compared with some typical algorithms, such as FS, TSS,NTSS and DS, MBC-ASW is improved on the compression performance and the efficiency, thus it is adaptive for different sequences.

关 键 词：运动估计 块特性 自适应搜索窗口 算法 

分 类 号：TN91[电子电信—通信与信息系统]