Video-based vehicle tracking based on moving object detection

Presented a robust and efficient video-based vehicle tracking method based on moving object detection to handle the dynamic illumination changes, shadow and occlusion in complex traffic scene. Firstly, an adaptive background modeling was adopted to detect the moving regions. Secondly, shadow reduction was used to obtain exact moving area. Then, an occlusion resolving strategy was proposed to handle the occlusion. Finally, vehicle tracking was achieved by region correspondence, combined with Kalman filters to model the motion of vehicles and optimize the vehicle parameters. The experimental results demonstrate that the proposed vehicle tracking method can effectively resolve shadow and occlusion problems in complex and variable outdoor scenarios, and robust vehicle tracking can be achieved.