雷达科学与技术2012，Vol.10Issue(6)：618-623,6.
基于稀疏约束的SAR目标特征提取方法研究
SAR Target Feature Extraction Based on Sparse Constraint Nonnegative Matrix Factorization
摘要
Abstract
Feature extraction is a key technology and the core task in synthetic aperture radar(SAR) target recognition. A non-negative matrix decomposition of the image target feature extraction method based on sparseness constraint is presented in this paper. In order to improve the similarity within the class and the differences between the classes of feature vectors, the SAR target image is decomposed by use of sparseness constraint non-negative matrix factorization method to build sparse target feature vector. MSTAR data are used for target identification test based on support vector machine classification. The results show that the proposed method improves the stability and accuracy of target recognition significantly compared with PCA, ICA and general NMF characteristics extraction methods.关键词
合成孔径雷达(SAR)/非负矩阵分解/稀疏约束/分段光滑约束函数/支持向量机Key words
synthetic aperture radar(SAR)/ non-negative matrix decomposition/ sparseness constraint/ adaptive function/ support vector machines分类
信息技术与安全科学引用本文复制引用
高馨,曹宗杰..基于稀疏约束的SAR目标特征提取方法研究[J].雷达科学与技术,2012,10(6):618-623,6.基金项目
中央大学基础研究基金(No.ZYGX2009Z005) （No.ZYGX2009Z005）
