Detection and Tracking of Vehicles Based on Video and 2D Radar Information

For practical application, over-speed behavior of a vehicle is generally detected by the combination of the radar speedometer and the camera capture. Little information, just a single image of the vehicle and the velocity is recorded as the evidence. If other vehicles are in the sight of the camera, this method cannot identify which one the measured velocity belongs to. The velocity is even inaccurate due to the influence. Focusing on this problem, we introduce a detection and tracking system of vehicles based on the combination of the video and 2D radar information. Multiple vehicles in the sight of the camera are detected and tracked, whose radar information is calibrated and embedded into the image coordinate system. The trajectory and velocity of each vehicle is recorded, and the over-speed judgment of multi-target can be solved. The system is evaluated in the experiments and the practical application. The results show that the proposed fusion algorithm of the video and radar information has a good performance on the detection, tracking and velocity measurement of vehicles.