基于样本紧密度的雷达高分辨距离像识别方法研究  

Radar High Resolution Range Profiles Recognition Based on the Affinity

作　　者：张瑞[1,2] 牛威[1,2] 寇鹏[1,2] 

机构地区：[1]宇航动力学国家重点实验室,西安710043 [2]西安卫星测控中心,西安710043

出　　处：《电子与信息学报》2014年第3期529-536,共8页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(61179010)资助课题

摘　　要：模板库是基于高分辨距离像(HRRP)进行雷达自动目标识别的关键。由于真实目标复杂的电磁散射特性及背景杂波的存在,基于传统平均HRRP模板库的目标识别方法的实用性能有限。该文分析了雷达目标HRRP样本集在单位超球面上的空间分布特征,构建了"类心+紧密度球"特征模板库,定义了基于紧密度球的距离度量方法,最后提出基于样本紧密度的HRRP识别方法。基于公共MSTAR数据的实验结果表明该方法具有良好的工程实用性能。Template database is the key to radar automation target recognition based on High Resolution Range Profile （HRRP）. Because of the complicate electromagnetic scattering in real target and background clutter, the traditional method based on the template of average HRRP does not has a good performance sometimes. The geometric structure of HRRPs on the unit hypersphere is analyzed, and the templates of cluster center and affinity sphere are built. Based on the affinity sphere, the new distance measure is defined. Finally a new recognition method is proposed based on the affinity among HRRPs. The experiment results based on MSTAR data demonstrate that the proposed method achieves a good engineering performance.

关 键 词：雷达自动目标识别 高分辨距离像 姿态敏感性 特征提取 紧密度 

分 类 号：TN957.51[电子电信—信号与信息处理]