基于瞬态极化时频分布及奇异值特征提取的雷达目标识别  

Radar Target Recognition Based on Instantaneous Polarization Time-Frequency Distribution and Singular Values Feature Extraction

作　　者：曾勇虎[1] 王雪松[1] 肖顺平[1] 庄钊文[1] 

机构地区：[1]国防科技大学电子科学与工程学院

出　　处：《电子学报》2005年第3期571-573,共3页Acta Electronica Sinica

基　　金：国家杰出青年科学基金 (No.60 0 2 51 0 2 ) ;全国优秀博士论文专项资金 (No .0 81 0 0 1 0 1 ) ;国防预研项目 (No .41 30 351 7)

摘　　要：本文介绍了瞬态极化时频分布的概念 ,并给出了瞬态极化Wigner -Ville分布及伪Wigner -Ville分布的定义 ,用以刻画雷达目标回波在时频域上的极化特性 .然后 ,由回波的瞬态极化伪Wigner -Ville分布各分量的奇异值组成特征矢量 ,以BP神经网络作为分类器进行了目标识别实验 .实验结果表明 ,与传统的对目标极化回波的两个分量分别作时频分析的处理方法相比 ,文中方法具有更好的目标识别性能 .This paper introduces the concept of instantaneous polarization time-frequency distribution (InPol TFD), and presents the definition of instantaneous polarization Wigner-Ville distribution (WVD) and pseudo Wigner-Ville distribution (PWD), which are used to describe the polarization property of radar target s return in time-frequency domain. Next, a feature vector, formed by the singular values of each component of the InPol PWD of target's return, is used as input vector for the classifier based on BP neuron network. The target recognition results show that the proposed method has better performance than conventional one.

关 键 词：瞬态极化 时频分布 目标识别 

分 类 号：TN957[电子电信—信号与信息处理] TN959[电子电信—信息与通信工程]