Abstract: The paper proposes a multisensor track association algorithm based on the optimal subpattern assignment (OSPA) distance. The target track is described as a set. The OSPA distance is employed to evaluate the difference between any two track sets, where the tracks with the shortest OSPA distance belong to the same target track. Considering the real time requirement, a recursive algorithm with a sliding window is further proposed. The final results show that the proposed algorithm can keep track history information and track geometric features. Thus it can effectively deal with track cross, track split, and track asynchronous cases. Compared with the weighted track association approach and the independent sequential approach, the proposed algorithm performs better in associating rates and asynchronous problems.

Key words: tracking, track association, OSPA distance, track set, track cross, split and asynchronous, fixed sliding window