Journal of Systems Engineering and Electronics

• 电子技术 •    下一篇

基于距离-方位二维聚类的海上编队目标距离像提取

罗小波1,2,范红旗1,宋志勇1,付强1   

  1. 1. 国防科学技术大学电子科学与工程学院ATR 重点实验室,湖南 长沙 410073;
    2. 中国人民解放军96363 部队,甘肃 天水 741020
  • 收稿日期:2012-06-14 修回日期:2013-01-30 出版日期:2013-07-22 发布日期:2013-05-15

Range profiles extraction for maritime formation targets based on range-azimuth clustering

LUO Xiaobo1,2,FAN Hongqi1,SONG Zhiyong1,FU Qiang1   

  1. 1. ATR Key Laboratory, School of Electronic Science and Engineering, National University of Defense
    Technology, Changsha 410073, China; 2. Unit 96363 of the PLA, Tianshui 741020, China
  • Received:2012-06-14 Revised:2013-01-30 Online:2013-07-22 Published:2013-05-15

摘要:

在观测编队目标时波束内容易出现多个目标,可能引起距离像邻近甚至混叠,难以判断距离像簇与目标之间的对应关系,导致无法准确提取单个目标的距离像信息。本文基于距离像目标识别的需求,在海面编队结构信息的约束下,分析了波束内多目标出现概率,定义了多目标距离像之间的4 种关系,利用目标的距离和方位角信息进行二维聚类来判断距离像簇与多目标之间的对应关系,从而实现波束内多个舰船目标的距离像提取。仿真实验表明该算法能够在不同信噪比和带宽条件下很好地区分编队内不同目标的距离像,为下一步基于距离像信息的目标识别和稳定跟踪提供了重要条件。

Abstract:

When missile-borne radar observing maritime formation targets, multiple range profiles are likely vicinity even overlapped as a consequence of the existence of multiple targets in the antenna beam. It is difficult to judge the corresponding relation between range profile cluster and target, which lead accurate extraction the single target range profile cannot implement. Based on the need of the range profile recognition and the constraints imposed by the target formation structure, a range profile extraction algorithm is proposed based on the range-azimuth clustering. The probability that multiple targets exist within the same beam and the four relationship between multiple targets range profiles are firstly analyzed and defined. Subsequently, the relationship between range profile cluster and multiple target are judge with clustering the range and azimuth information of the surface targets in the range-azimuth space. The algorithm finally extracts the range profiles of each individual ship target within the beam according to the clustering information. Simulation results demonstrate that the algorithm performs well in distinguishing different targets range profiles under a wide range of signal-to-noise ratio (SNR) and bandwidth, meaning that it can provide a solid foundation for the subsequent target recognition and tracking applications.