石绍应　杜鹏飞　张靖　曹晨

(1.中国电子科学研究院,北京 100041;2.空军预警学院,武汉 430019)



基于雷达测量的多目标联合检测、跟踪与分类方法

石绍应1,2杜鹏飞2张靖1曹晨1

(1.中国电子科学研究院,北京 100041;2.空军预警学院,武汉 430019)

摘要利用雷达测量中的目标速度、加速度等属性信息,基于跳转马尔科夫系统模型高斯混合概率假设密度滤波算法,提出了一种多目标联合检测、跟踪与分类方法. 该方法在进行雷达多目标测量信息处理的多模型混合高斯概率假设密度滤波过程中,对各高斯项编号,进行航迹提取,在滤波处理的同时形成带有航迹编号的明确航迹,并进行航迹管理;同时,根据目标运动模型,联合利用目标加速度控制输入与速度估计进行多目标分类. 仿真试验验证了该方法能够在检测、跟踪的同时,对目标航迹进行有效类型识别.

关键词多运动模型;多目标联合检测、跟踪与分类;高斯混合概率假设密度滤波;航迹管理

DOI10.13443/j.cjors.2015040202

Multi-target joint detection, tracking and classification using radar information

SHI Shaoying1,2DU Pengfei2ZHANG Jing1CAO Chen1

(1.ChinaAcademyofElectronicsandInformationTechnology,Beijing100041,China;2.AirForceEarlyWarningAcademy,Wuhan430019,China)

Abstract Based on the jump Markov system model Gaussian mixture probability hypothesis density filtering (JMS-GMPHDF), a method is proposed for multi-target joint detection, tracking and classification by using the kinematic information of radar targets such as velocity and acceleration. This method applies track extraction technique and assigns tag to each Gaussian item during multi-model Gaussian mixture probability hypothesis density filtering in the process of radar multi-target measured information, which can form and manage the clear track with tracking number. Meanwhile, based on target kinematic models, the multi-target are classified by jointly using acceleration input control and velocity estimation of the target. Simulation results suggest that the proposed method can classify the target track effectively during the detection and tracking.

Keywords multiple kinematic model; multi-target joint detection; tracking and classification; Gaussian mixture probability hypothesis density filtering; track management

引言

在现代战场环境中,预警监视系统需要快速、准确地判别目标是否存在以及目标数量,估计目标的状态参数,识别目标的类型、型号、敌我属性,判断目标的作战意图. 这包括预警监视系统对目标的检测、跟踪、识别、态势估计与威胁评估全过程,一般需要雷达、ESM/CSM、IFF、红外等多传感器协同完成. 然而,单就雷达而言,其获取的测量信息除了通常情况下用于目标检测、跟踪外,其中包含的目标飞行包线、目标机动能力等信息可用于目标……