在压缩域内直接分割运动对象对于有实时要求的应用而言是十分必要的，H．264以其优越的压缩效率已经在许多应用中逐渐取代了MPEG-2／4，但有关在H．264压缩域内进行运动对象分割的研究还很少。为此提出了一种从H.264压缩域实时分割运动对象的算法，该算法首先对从H．264视频中提取出的原始运动矢量场进行时域和空域的归一化，接着通过对连续多帧的运动矢量场进行累积来增强显著的运动信息；然后对累积运动矢量场进行全局运动补偿，同时利用快速的统计区域生长算法按照运动相似性将其分割成多个区域；最后利用运动矢量场的方向角直方图来判断出属于运动对象的分割区域，以组成运动对象。通过对多个MPEG-4测试序列的实验结果表明，该方法不仅能够从H．264压缩域中实时地分割出运动对象，且具有良好的分割质量。

Moving object segmentation in the compressed domain is absolutely necessary for real-time applications.Due to the predominant compression efficiency,the emerging video coding standard H.264 is replacing MPEG-2/4 in many multimedia applications,but moving object segmentation in the H.264 compressed domain is rarely investigated until now.In this paper,we present a new approach to segment moving objects from the H.264 compressed domain.The motion vector(MV) field extracted from the H.264 compressed video is first normalized in both temporal and spatial domain,and the MV fields of several continuous frames are accumulated to enhance the salient motion.Then the global motion compensation is performed on the accumulated MV field,while the fast statistical region growing algorithm is exploited to segment it into different motion-homogenous regions.Finally,the orientation histogram of the MV field is exploited to determine the moving object regions.Experimental results for several MPEG-4 test sequences demonstrate the proposed approach can segment moving objects from the H.264 compressed domain in real time,and exhibit good segmentation quality.