利用H.264中运动矢量实现运动目标检测

Moving object detecting technique based on motion vector in H.264 coding

针对视频压缩和运动目标跟踪同时实现的应用,通过研究H.264压缩标准中运动矢量包含的图像运动信息并分析运动估计原理,提出了一种利用视频压缩中的运动矢量信息实现运动目标检测的方法,确立了H.264编码流中运动矢量与场景中物体运动状态的对应关系.将运动目标从背景中分离是检测算法的核心.对于双门限值的设置,可分离不同运动速度的目标;同时,算法排除了背景运动的干扰,因而可应用于摄像机运动的场合;由于检测算法所用的运动矢量直接来源于H.264编码过程,而大大降低了计算复杂度,利于硬件实现.

Considering the applications of joint realization of video compression and moving object tracking, by studying the moving information included in the motion vector of the H. 264 video coding international standard and analyzing motion estimation principle, a motion vector based on moving object detecting technique was proposed, and the relationship between the motion vector and the moving object was set up. The key part of detecting algorithm is to divide moving object from background. By setting two threshes properly, the detected objects could be changed according to their moving velocity. As the disturbance of background moving eliminated by the algorithm, it can be used in some cases when the camera is moving. For motion vectors come from H. 264 coding progress directly, the calculation complexity is reduced, so it is easy to realize by hardware.