Study Of Methods For Moving Human Detection And Tracking

The detection and tracking of moving human is an important aspect in the videoanalysis and a valuable field in computer vision as well. This paper mainly discusses thedetection and tracking of moving human in video sequences gathered from a fixedcamera. As the existing mixture gaussian models could not solve the problems such asghost and moving shadow detection, an improved mixture model updating method isproposed, Which increases the speed of background modeling and eliminates the ghostquickly; A shadow detection scheme based on both tricolor coefficient invariance andluminance change detection is proposed. Based on moving objects detection, this paperpresents a moving object tracking method, which use outside rectangular of movingobjects and color template to track object in the scene. At last, the theory how activeshape model applied to contour extraction is analyzed and the active shape model basedon color information presented with a great improvement of the accuracy of contourextraction in video.