基于HRRP识别信息辅助多目标跟踪的仿真研究

Simulation of Classification-Aided Tracking Multiple Targets Based on HRRP

密集多回波环境下对弹道多目标跟踪时,融合目标运动学信息和HRRP分类信息,研究了基于HRRP分类信息辅助跟踪的最近邻算法CATNN;由于受获取实际数据的限制,为检验算法,结合目标的仿真数据,提出了多目标跟踪和识别动态交互的仿真方法;在建立的场景中与常规的JPDA、NN跟踪算法进行比较,结果表明CATNN克服了轨迹合并和误跟的现象,具有较高的多目标跟踪性能,且产生的仿真数据满足了CATNN算法验证的需求。

In ballistic multiple targets tracking in dense clutter environment, a novel CATNN （Classification-Aided Tracking Nearest Neighbor） algorithm was investigated which fused kinematical information and classifying information based on HRRP. Because of the difficulty in obtaining the real data, a simulative method of dynamically combining tracking and classification was put forward with static simulation data to evaluate the algorithm. In the simulation scene, the performance of CATNN was compared with conditional JPDA and NN algorithm. The results indicate that the CATNN can overcome the track-coalition and false-tracking phenomena, and the demand of algorithm verification can be satisfied with these simulation data.