The background model initialization algorithm with iteration form is presented through doing the research on the features of background model and foreground detection in YUV space in the paper. It also describes background subtraction technique to detect the motion objects, and provides a well and efficient human recognition algorithm by employing the unique features of human figure and the priori knowledge of the scene. In order to track the motion state of human in image sequence accurately, we make use of combined algorithm to prevent Kalman filter from diverge. An automatic human detection and tracking system was formed in the paper, which could analyze and process the image sequence gathered from one fixed camera, detect and keep tracking of human motion. The system has been demonstrated through a series of tests with image sequences gathered from both indoor and outdoor, showing great practicability. |