Multi-Target Association in the Distributed Passive Sensor System with Clustering-Based Fusion and Association Capability Analysis

Accurate multi-target association in distributed passive sensor systems is crucial for collaborative perception, especially in dense target scenarios that often produce false positives. This paper proposed an improved approach for multi-target association in the distributed passive sensor system with clustering-based fusion and association capability analysis. We first associate targets based on the perpendicular foot distance of the lines of sight from dual sensors to the targets, considering the tracking information of multiple continuous frames in 2D images. The selected primary sensor will perform target association with the other auxiliary sensors, respectively. Association capability analysis provides a suggestion for the sensor formation setting, effectively reducing false positives. Finally, a clustering algorithm is used to refine the overall association results. Experimental results demonstrate that our proposed method outperforms the baseline method in terms of association precision, particularly under varying settings of sensor errors and target densities.