Research on MPEG Motion Vector Based Automatic Tracking Systems Using Network Cameras

在本课题中，我们从事的是利用网络摄像机进行自动目标跟踪的研究。我们研究了利用网络摄像机的MPEG视频流进行实时运动分析和跟踪控制的基本技术。本文提出了一种运动目标自动检测与跟踪方法和一种新的运动矢量噪声消除方法，这两种方法是利用MPEG运动矢量进行视频处理必不可少的组成技术。通过对实际MPEG视频数据的实验验证了该方法的有效性。

In this research subject, we engaged in developing automatic object tracking using network cameras. We studied fundamental techniques for real-time motion analysis and tracking control using MPEG video streaming of network cameras. We developed an automatic motion object detection and tracking method and a novel noisy motion vector elimination method that are indispensable component technologies for video processing using MPEG motion vectors. Their effectiveness was demonstrated through several experiments using actual MPEG video data.