Occlusion adaptive object tracking based on video image and radar data

Vehicle and pedestrian detection is important in traffic surveillance, and video image processing technology has evolved and offered promising prospects. Occlusion object tracking is difficult in image processing, and this paper gives a system using video image and radar data to segment the occlusion objects. In this system the objects are extracted by processing image sequence, and using the radar data for object tracking when there is an occlusion. First, in this paper, an improved method is proposed for background acquisition before object extraction and location, and then an algorithm is adopted for shadow clutter rejection. Second, object tracking with image and radar data is discussed. The experimental results show that this system can robustly track objects, and significantly improve the accuracy probability of tracking with occlusions.