基于LDLPP的SAR目标型号识别

刘明; 陈士超; 武杰; 卢福刚; 邢孟道

引用本文:	刘明，陈士超，武杰，卢福刚，邢孟道. 基于LDLPP的SAR目标型号识别[J]. 雷达科学与技术, 2018, 16(4): 439-445.[点击复制]
	LIU Ming, CHEN Shichao, WU Jie, LU Fugang, XING Mengdao. SAR Target Configuration Recognition via LDLPP [J]. Radar Science and Technology, 2018, 16(4): 439-445.[点击复制]

【打印本页】【下载PDF全文】【查看/发表评论】【下载PDF阅读器】【关闭】

←前一篇|后一篇→

过刊浏览高级检索

本文已被：浏览 6061次下载 922次	码上扫一扫！
分享到：微信更多字体:加大+\|默认\|缩小-
基于LDLPP的SAR目标型号识别
刘明，陈士超，武杰，卢福刚，邢孟道
1.陕西师范大学计算机科学学院，陕西西安710119;2.中国兵器工业第203研究所，陕西西安710065;3.西安电子科技大学雷达信号处理国家重点实验室，陕西西安710071

摘要:

局部保持投影算法可保持数据的局部结构，但其无法实现异类相似样本的有效分离，针对此问题，提出了一种融合样本先验类别信息的局部保持投影算法（LDLPP），实现SAR目标型号识别。所提算法将样本的先验类别信息融入局部保持投影模型中，根据类别信息构造相似性矩阵和差异性矩阵以实现数据结构的有效捕获与保持，利用相似性矩阵保持降维前后同类样本之间的局部结构，利用差异性矩阵扩大降维后异类相似样本彼此之间的距离。采用实测的MSTAR数据进行SAR目标的型号识别，实验结果验证了所提算法的有效性。

关键词: 局部保持投影类别信息 SAR图像目标型号识别

DOI：10.3969/j.issn.1672-2337.2018.04.015

分类号:TN957

基金项目:国家自然科学基金(No.61701289, 61601274)；中央高校基本科研业务费专项资金(No.GK201603089)

SAR Target Configuration Recognition via LDLPP

LIU Ming, CHEN Shichao, WU Jie, LU Fugang, XING Mengdao

1.School of Computer Science, Shaanxi Normal University, Xi'an 710119, China;2.No. 203 Research Institute of China Ordnance Industries, Xi'an 710065, China;3.National Key Laboratory of Radar Signal Processing, Xidian University, Xi'an 710071, China

Abstract:

Locality preserving projections (LPP) is a powerful tool to preserve the local structure of the datasets, but it lacks the ability to set the samples close to each other with different labels apart. In this paper, a label-dependent locality preserving projections (LDLPP) algorithm is presented to ease the problem for SAR target configuration recognition. Two structure preserving matrices are constructed based on the label information to realize intrinsic structure preserving of the datasets. The similarity matrix is constructed to preserve the local structure of the samples with the same label, whereas the divergence matrix is utilized to set apart the samples close to each other with different labels. The advantage of the proposed algorithm is verified on the moving and stationary target acquisition and recognition (MSTAR) database with extensive experiments.

Key words: locality preserving projections (LPP) label information SAR images target configuration recognition