Studies On Detection & Tracking Of Moving Human In Video Sequence

Recently as the basis of recognition in machine vision field, Real-time object tracking becoming more and more important in the video analysis and processing field. Especially the detection and tracking of the moving human target has more extensive applicable value under the environment of the global safety problem. The safety surveillance system based on the computer vision can both finish the job efficiently and save the cost of human resource and property.Moving human target include two parts, which are partial moving of the human body and entire human body moving. In this paper we illustrate the arithmetic used on both parts. For the detection and tracking of partial human body moving, we mainly do the research on the real time hand contour tracking. We present a new approach to real-time hand contour tracking from color video sequences. It represents the shape in a single frame by means of improved radius-vector functions, combining splitting techniques and merging techniques. Tradition kalman filter is also used to estimate the hand motion track between the neighbor frames. It has proved that it can finely overcome its shortcoming. It can track the motion hand or the transform of the hand gesture robustly.For the detection and tracking of the entire human body, firstly we do the background re-construction arithmetic to reconstruct the background. Then we use the background subtraction arithmetic to detect the moving human body. Secondly we process the corner detection arithmetic to detect the corners on the human body, then determine the rectangle outside the human body and make it as the initial contour of the SNAKE model. Through this way we can realize the auto scaling the SNAKE contour. Make the contour in the right frame as the initial SNAKE contour of the next frame, so can complete the tracking work. However, for the sake of the convergence of the SNAKE depends much on the initial contour of the snake, we present a optical flow arithmetic to estimate the displacement of the contour points between frames so that make our arithmetic more robust.