摘要
SVM是解决非线性图形识别问题非常有效的分类方法。本文提出了一种SVDA分类方法,充分利用了SVM的内在优良推广能力。通过寻找有限样本情况下的最优分类面法线方向作为投影轴,对样本数据进行投影,提取样本的特征,进而实现目标识别。本文将SVDA分类方法应用于MSTAR数据集进行SAR雷达目标识别实验,得到了较好的识别效果。
SVM（ support vector machine） is an effective classifying method to solve non- linear pattern recognition. A classifying method called SVDA（ Support Vector Discriminant Analysis,） is presented,which makes full use of intrinsic perfect generalization ability of SVM. The sample data is projected to extract characteristic of the sample so as to implement target recognition by looking for normal direction of optimal hyperplane in condition of limited condition as projection axis. SVDA classifying method is applied in MSTAR data set to perform SAR radar target recognition test,and the better recognition effect is obtained.
出处
《火控雷达技术》
2015年第4期5-7,19,共4页
Fire Control Radar Technology
关键词
支持向量机
特征提取
分类器
雷达目标
support vector machine
feature extraction
classifier
radar target
