摘要
研究了一种基于前向散射雷达的车辆目标自动识别方法:分析了前向散射雷达回波与目标速度、轮廓等目标特性之间的关系;提出了主瓣对齐的功率谱预处理方法,采用主成分分析方法分析了不同类别车辆的前向散射雷达功率谱的特性;提出了基于多元线性回归的目标特征提取方法,比较了K最近邻法和Bayes分类器对不同特征集的识别性能.实现结果表明,本文采用的特征提取方法显著改善了识别效果.
An automatic vehicle classification method based on forward scattering radar is proposed.The connections between the power spectrum with the silhouette and the speed of the target are analyzed.The power spectrum of several types of cars is preprocessed using a novel main lobe alignment approach,and then analyzed using the PCA method.The classification performance of the K-nearest neighbor and Bayes classifiers are compared on the feature extracted from the power spectrum using multiple linear regression.Experiment results demonstrated that the proposed method improves the classification performance significantly.
出处
《中国科学：信息科学》
CSCD
2012年第11期1471-1480,共10页
Scientia Sinica(Informationis)
基金
国家自然科学基金(批准号:61172177
61101229
61032009)资助项目
关键词
前向散射雷达
自动车辆识别
多元线性回归
BAYES分类器
forward scattering radar
automatic vehicle classification
multiple linear regression
Bayes classifier
