一种基于MPEG压缩域的运动对象分割算法  

New Moving Object Segmentation Algorithm in MPEG Compressed Domain

作　　者：刘龙[1] 刘贵忠[1] 刘洁瑜[1] 王占辉[1] 

机构地区：[1]西安交通大学电子与信息工程学院,西安710049

出　　处：《西安交通大学学报》2004年第12期1264-1267,共4页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目 (6 0 2 72 0 72 ) ;教育部跨世纪优秀人才基金资助项目 ;教育部"十五""2 11"工程西安交通大学重点科研项目

摘　　要：针对目前视频压缩域中运动对象分割准确率不高的问题 ,提出一种有效的压缩域运动对象的分割算法 .该算法先对运动矢量进行运动累加和中值滤波处理 ,然后利用离散余弦变换 (DCT)系数块之间的相关性对非零运动矢量区域进行校正 .由于在对运动矢量处理的同时加入了空间DCT块的相关性校正 ,因此分割准确率有了较大的提高 .实验统计结果表明 ,对几种不同视频场景中的运动对象进行分割 ,其分割率比现有的算法提高了约 15 % .Focusing on the problem that there is low segmentation veracity for moving objects in the video compressed domain, an effective segmentation algorithm was proposed, in which the motion vectors were processed by motive accumulation and median filter, and then the areas of non-zero motion vectors were corrected by using the relativity among discrete cosine transform (DCT) coefficient blocks. Because of adding the relativity correction of spatial DCT blocks while motion vectors were processing, the segmentation veracity is improved greatly. For various video scenes, the experiments results show that the segmentation veracity of the proposed algorithm is improved about 15% compared to that of the other algorithms.

关 键 词：分割 压缩域 运动对象 

分 类 号：TN911.73[电子电信—通信与信息系统]