Improved Tracking Algorithm for Multiple Targets
Sheng-Yun Hou, Chenwei Wu, Hui-Feng Hung, Shun‐Hsyung Chang · 2008
In this paper, a novel algorithm for tracking multiple targets is proposed. The algorithm incorporates subspace tracking with Kalman filtering to improve tracking performance. At each recursive step of the Kalman filter, subspace tracking is first combined with the MUSIC spectrum to provide angle estimates which are used as measurement data. To avoid abnormally large errors caused by crossing trajectory, the estimated angles are updated through a decision mechanism. Finally, the angle estimates are smoothed and then predicted for accuracy improvement and correct association, respectively. Simulation results are furnished to demonstrate the effectiveness of the proposed algorithm.