摘要
提出了基于零空间的线性直接辨别分析与非线性推广直接辨别分析方法,并将其用于雷达目标一维距离像识别。与传统子空间方法相比,上述两种方法保留并充分利用了类内散度矩阵最具分辨力的零空间信息,因而大大提高了目标的识别性能。对三种实测飞机数据的识别结果表明了所提方法的有效性。
Two null space-based algorithms, direct discriminant analysis (DDA) and its nonlinear extension generalized direct discriminant analysis (GDDA), are proposed and used in radar target recognition based on one-dimensional range profile. Comparing with other traditional subspace methods, DDA and GDDA hold and take full advantage of the null space of within-class scatter matrix, which has been demonstrated to contain the most important discriminant information. Therefore, the classification performance is improved significantly. Experimental results based on three measured airplanes data have confirmed the effectiveness of the proposed methods.
出处
《电波科学学报》
EI
CSCD
北大核心
2007年第6期1020-1024,共5页
Chinese Journal of Radio Science
基金
国家自然科学基金(No60372022)
新世纪优秀人才支持计划(NCET-05-0806)资助
关键词
雷达目标识别
一维距离像
直接辨别分析
核方法
特征提取
radar target recognition, range profile, direct discriminant analysis, kernel-based methods, feature extraction