A Novel Multi-vehicle Tracking Method Based on Linked List of Associated Vehicles
Hongtao Wu, Ying Meng · 2019
Vehicle tracking is a hot research direction in the Intelligent Transportation System (ITS). To deal with the traditional Camshift algorithm can not achieve multi-vehicle tracking, and tracking-lost of the vehicle under large proportion occlusion, a novel multi-vehicle tracking method based on linked list of associated vehicles is proposed in this paper. In order to initialize the search window, vehicle detection is used to obtain the target vehicle's location and size. Each tracked vehicle establishes a track list to realize multi-vehicle tracking. The algorithm uses Bhattacharyya distance to occlusion identification, and uses Kalman filter to predict the location of vehicles, accurate vehicle tracking can be realized under occlusion. Experimental results show that the algorithm can achieve a robust and accurate vehicle tracking under different traffic.