基于协同探测数据融合的水下多目标跟踪

Underwater Multi-Target Tracking Based on Collaborative Detection Data Fusion

在高杂波密度的水下环境中,若只利用局部传感器来跟踪多个目标,会产生较高的虚警,而且跟踪效果受限于传感器本身性能。为此文中基于分布式数据融合模型,首先在各个无人平台上利用联合概率数据关联(JPDA)形成局部航迹,然后再用匈牙利算法实现对关联矩阵的最优分配从而达到航迹的实时关联,最后通过凸组合融合算法形成全局航迹。仿真结果表明,经过此方法完成的多目标跟踪结果具有更好的可靠性,其跟踪精度基本在两站跟踪精度之间且具有较低的跟踪误差,适合在未知基站估计精度的系统中获得更为可信的估计结果。

In the underwater environment with high clutter density,high rate of false alarms will be generated if only local sensors are used to track multiple targets,meanwhile the tracking effect is limited by the performance of the sensors.Therefore,based on the distributed data fusion model,this paper first uses the joint probability data association(JPDA)to form a local track on each unmanned platform,then uses Hungarian algorithm to implement the optimal allocation of the correlation matrix so as to achieve real-time track correlation,finally forms a global track through the convex combination fusion algorithm.Simulation results show that the multi-target tracking results of this method are more reliable,and the tracking accuracy is almost between the tracking accuracy of two stations with smaller tracking error.It is concluded that the present method is suitable for obtaining more reliable estimation results in the system with unknown base station estimation accuracy.