一种超宽带SAR地雷散射方位不变特征提取新方法  

A Novel Ultra-Wide Band SAR Landmine Scattering Aspect Invariance Feature Extraction Method

作　　者：金添[1] 周智敏[1] 

机构地区：[1]国防科学技术大学电子科学与工程学院,湖南长沙410073

出　　处：《电子学报》2007年第12期2272-2278,共7页Acta Electronica Sinica

基　　金：国防科技大学优秀研究生创新资助项目(No.20050401)

摘　　要：利用机载或车载超宽带合成孔径雷达(UWB SAR)可以实现大区域地雷和雷场远距离快速探测,是探雷的发展趋势.为了降低地雷检测中的虚警,需要提取有效的地雷特征区分地雷和杂波.本文提出了一种基于高维时频分布(HDTFD)的地雷散射方位不变特征提取方法,具体实现可采用高维Wigner-Ville分布(HDWVD)和高维Choi-Williams分布(HDCWD)等.基于HDTFD特征提取方法能够在保持高空间分辨的同时,有效提取散射函数中关于频率和方位角信息.利用Rail-GPSAR系统实测数据比较了基于HDWVD和基于HDCWD特征提取方法在分辨率、交叉项抑制和方位不变特征提取等方面的性能.实测数据处理结果说明基于HDCWD特征提取方法更适合实际数据处理.Air-or vehicle-borne Ultra-Wide Band Synthetic Aperture Radar （UWB SAR） has the capability to detect landmines and landmine fields over large area from a standoff distance quickly, which is thetrend of landmine detection.In order to reduce the false alarms in landmine detection,it is needed to extract efficient landmine feature to classify landmine and clutter.In this paper, a High-Dimensional Time-Frequency Distribution （HDTFD） based method to extract the aspect invariance feature of landmine scattering is proposed, where the realization can be performed with High-Dimensional Wigner-Ville Distribution （HDWVD） and High-Dimensional Choi-Williams Distribution （HDCWD）. The proposed HDTFD based feature extraction method can extract the frequency and aspect angle dependent information in scattering function efficiently while maintains high spatial resolutions. The HDWVD and HDCWD based feature extraction methods have been compared using the real data collected by the Rall-GPSAR system in terms in resolution, cross-terms suppression and aspect invariance feature extraction. Field data processing results have shown that the HDCWD based feature extraction method is more suitable for real data processing.

关 键 词：超宽带 合成孔径雷达 特征提取 高维时频分布 地雷 

分 类 号：TN957[电子电信—信号与信息处理]