Research on traffic moving object detection, tracking and track-generating
Shanming Lin, Jun Tang, Xuewu Zhang, Yan-yun Lv · 2009
This paper presents algorithms for vision-based detection, tracking and trajectory generation of vehicles in simple video sequences of traffic scenes which are registered by a stationary camera. Processing is done by three steps: improved vehicle detection method, faster tracking algorithm and new vehicle trajectory generation technique. Vehicles are viewed as rectangular blocks with certain active action. The propounded method is based on the correspondences between regions and vehicles, as the vehicles moving through the video sequence. The vehicle trajectory generation technique, used for gathering statistical traffic data for analysis in vehicle flow detection, is also described in detail. Experimental results from highway scenes demonstrate the effectiveness of the vehicle detection and tracking method.