基于支持向量机的浅地层探地雷达目标分类识别研究  

Research on Ground Penetrating Radar Target Identification Based on Support Vector Machines in Shallow Subsurface Application

作　　者：张春城[1] 周正欧[1] 

机构地区：[1]电子科技大学电子工程学院704教研室,四川成都610054

出　　处：《电子学报》2005年第6期1091-1094,共4页Acta Electronica Sinica

摘　　要：对于探地雷达用于探测地雷情况下的目标识别,从实用性上说只需识别地雷或非地雷两类目标即可,而通常的支持向量机正是用于两类分类.本文结合浅地层探地雷达数据的特点,提出了一种基于支持向量机的浅地层探地雷达目标分类识别方法,并分析了时域,傅立叶谱,及离散小波变换三种特征数据用于所提方法时的效果.通过对实测数据进行处理,结果表明三种特征数据用于所提方法都能取得较好的效果.Ground penetrating radar (GPR) is an effective tool to detect mine. Ground penetrating radar target identification needs to distinguish mine and non-mine only, and support vector machines (SVM) can be used to two-class classification problem in practice. This paper analyzed the characteristic of shallow subsurface GPR data, proposed a method of GPR target identification based on support vector machines, and processed measurement data by using the proposed method in time sample, Fourier spectra, and discrete wavelet transform as feature data. The results of processing measurement data show this method is effective.

关 键 词：探地雷达 支持向量机 目标识别 地雷 

分 类 号：TN955.1[电子电信—信号与信息处理]