Multi-target tracking method based on dynamic clustering and two-way matching

To address the challenges of target trajectory tracking in multi-platform (multi-radar, multi-sonar) collaborative surveillance under complex maritime conditions, this paper proposes a multi-source data cooperative tracking method that integrates dynamic noise-resistant DBSCAN clustering, bidirectional verification Hungarian matching, and dynamic extended Kalman filtering. First, a dynamic noise-resistant DBSCAN clustering model is constructed to suppress heterogeneous noise and effectively eliminate noise points. A bidirectional verification Hungarian matching algorithm is proposed, which jointly optimizes forward and reverse cost matrices to significantly reduce the ID switch rate in dense target scenarios. Finally, a closed-loop feedback mechanism driven by a dynamic extended Kalman filter (EKF) is established to achieve adaptive optimization of parameters throughout the entire workflow, from data acquisition to trajectory prediction. Simulation experiments validate the effectiveness of the proposed method, which provides high-precision and robust technical support for maritime multi-domain collaborative surveillance, significantly enhancing global situational awareness of surface-air-underwater heterogeneous targets in complex sea conditions.