3D Multitarget Tracking Track Segment Association Method

In a complex battlefield environment,missing measurements,maneuvering of targets,and large tracking errors may lead to track breakages.A large number of track segments not only reduces track continuity and causes data redundancy in the tracking system,but also affects its situational awareness and resource allocation.This makes the multitarget tracking a serious challenge.Track segment association(TSA)algorithm solves the problem of track breakage by associating track segments of the same target,which has important research significance in the field of multitarget tracking.Traditional TSA is a batch algorithm,which is suitable for scenarios where only two track segments are yielded by one target.In 3D tracking,traditional TSA algorithm cannot achieve satisfactory association effect when faced with frequent track breakages caused by poor elevation precision.In view of the above situation,this paper studies the TSA problem of 3D multitarget tracking.The main work is as follows:1.To solve the TSA problem of 3D multitarget tracking,a 3D batch TSA algorithm with improved track fusion method is proposed.Under the framework of traditional TSA,the old track segment is predicted to the starting time of the paired young track segment,and the young track segment is backtracked to the termination time of the old track segment.Then the information matrix fusion method is utilized to fuse the track segments with memory in broken period.Compared with the traditional polynomial fitting method,the proposed algorithm can reduce the position estimate error of the target in broken period and improve the tracking accuracy.2.To deal with the TSA problem with poor elevation precision of the sensor,a3 D recursive TSA algorithm is proposed.When a target is tracked in 3D space with poor elevation precision,track breakages occur more frequent.In this case,a recursive track segment classification strategy is proposed to associate each track segment to the same target.Secondly,based on the historical velocity information of targets,the actual termination time of the old track segment and the actual starting time of the young track segment are recomputed.The inaccurate estimates are released to obtain a more accurate association result.The simulation results show that the proposed algorithm can effectively deal with the TSA problem with poor elevation precision and improve the association rate.3.Aiming at the TSA problem when the target motion model is mismatched with the assumed model and the elevation precision is poor,a 3D recursive TSA algorithm based on recursive Gaussian process is proposed.Firstly,the recursive classification method is still utilized.Secondly,the recursive Gaussian process is used to smooth the track segments to reduce the influence of estimate error,and the smoothed track segments are used as the training set of Gaussian process.Thirdly,the track prediction and backtracking in broken period are performed based on the motion model learned by the Gaussian process.Finally,the association is completed through the optimal assignment.The simulation results show that the proposed algorithm can improve the correct association rate of track segments when the assumed model and the target model are mismatched,and improve the track continuity.