Journal of Systems Engineering and Electronics ›› 2012, Vol. 34 ›› Issue (2): 263-269.doi: 10.3969/j.issn.1001-06X.2012.02.09

• 传感器与信号处理 • 上一篇    下一篇

基于复值HRRP CICA特征的多方位SAR目标识别

张新征1, 黄培康2   

  1. 1. 重庆大学通信工程学院, 重庆 400044;
    2. 中国航天科工集团科技委, 北京 100048
  • 出版日期:2012-02-15 发布日期:2010-01-03

Multi-aspect SAR target recognition based on features of sequential complex HRRP using CICA

ZHANG Xinzheng 1, HUANG Peikang2   

  1. 1. College of Communication Engineering, Chongqing University, Chongqing 400044, China; 2. The Science and Technology Committee, China Aerospace Science & Industry  Corporation, Beijing 100048, China
  • Online:2012-02-15 Published:2010-01-03

摘要:

提出了一种基于雷达目标复距离像复值独立分量分析(complex independent component analysis, CICA)的合成孔径雷达(synthetic aperture radar, SAR)目标多方位散射特征提取和识别方法。根据雷达目标散射机理,将目标高分辨率复距离像建模为多个散射中心的复相干叠加。在分析复距离像的基础上,采用CICA方法实现了距离像中每个散射中心响应的分离。针对每个散射中心响应,利用高阶矩方法提取特征矢量。分类器基于隐马尔可夫模型(hidden Markov model, HMM)设计。采用美国运动和静止目标获取与识别(moving and stationary target acquistion and recognition, MSTAR)计划公开发布的目标实测数据进行算法实验,实验结果说明了提出方法具有较好的识别率。

Abstract:

A novel method of synthetic aperture radar (SAR) targets multi-aspect scattering feature extraction and recognition is proposed based on complex independent component analysis (CICA) of sequential high range resolution profiles (HRRP). According to the radar target scattering mechanism, the target HRRP is modeled as a complex linear coherent combination of multiple scattering centers response. The seperation of each scattering center response is performed using CICA on the basis of the analysis of a complex HRRP. 〖JP2〗For each scattering center response, feature vectors are extracted utilizing high order moments. Hidden Markov model (HMM) classifiers are designed for target recognition. The experiment results with moving and stationary target acquistion and recognition (MSTAR) data sets show that a good classification performance is obtained.