A new vehicle tracking method with region matching based on Kalman forecasting model

Aiming to the problems of vehicle´s vanishing in short time and completely occlusion in complex traffic scenes, we proposed a method with region matching based on Kalman forecasting model to forecast and track the vehicle´s moving state. Firstly, the observation parameters such as centroid and block size of moving vehicle region are abstracted, and region models can be built for every vehicle. Secondly, the region models can be forecast and updated with Kalman filter. Two region model matching criterions are built for accurately orientating and tracking the moving vehicles. Finally, the complete occlusion can be solved with reasoning, and the part occlusion can be eliminated with a separation line through the vehicle´s shape analysis. The experimental results show that the proposed method can reduce the searching range of vehicle matching, effectively forecast the vehicle´s position and solve the complete occlusion of moving vehicle.