Skip to Main Content
Recently, the video-based technologies play an important role in monitoring fields, including the detection, identification and tracking of moving object. This paper designs an effective pedestrian detection and tracking algorithm in the traffic environment. Firstly, extract the moving prospect from the background in the video images, using conventional frame difference algorithm. The moving targets extracted include pedestrians, vehicles, the shadow of vehicles and the shaking tree at the side of roads. Then we analyze the geometry features of moving targets. According to the prior knowledge of pedestrian shape, the pedestrians are separated from the moving targets. Finally, on the basis of pedestrian detection and identification, we propose a tracking algorithm based on a combination of particle filter and mean shift. Test results showed that the algorithm have better real-time ability and robustness.