Efficient Multiple Extended Target Tracking Based on Minimum Spanning Tree

In this paper, an efficient multiple extended target tracking algorithm is proposed to solve the problem that multi-extended target tracking based on RFS (Random Finite Set) are difficult to implement with high computational burden. Firstly, the measurements at each time step are partitioned based on the MST (Minimum Spanning Tree) of measurements, including the internal relationships between the measurement partition results. Then, these relationships are used to avoid redundant calculation for extended target tracking under RFS filter framework, which can optimize the computational efficiency of the state updating. Finally, the simulation results demonstrate that our proposed method significantly improves the state update efficiency for extended target tracking based on RFS. Specifically, compared to the ET-PHD-DP filter, the state updating runtime of our method using the ET-PHD filter is reduced by 76.5%. Similarly, for the ET-CBMeMBer-DP filter, the state update runtime of our method using the ET-CBMeMBer filter is reduced by 90.0%.