基于目标结构特征像的SVM目标识别方法  

SVM Target Identification Method based on Target's Structure Characteristic Profile

作　　者：陈怀民[1] 郑颖[1] 段晓军[1] 

机构地区：[1]西北工业大学,陕西西安710065

出　　处：《火力与指挥控制》2009年第3期59-62,共4页Fire Control & Command Control

基　　金：航空基金资助项目(20061353017)

摘　　要：针对支持向量机(SVM)对目标高分辨一维距离像(HRRP)识别率和稳定性不高的问题,研究将目标分解理论推广应用于对目标极化散射矩阵的分解,求得了目标结构特征像。在分析研究SVM的训练和测试方法的基础上,采用SVM通过结构特征像对两实验目标进行分类识别,结果表明,基于目标结构特征像的SVM目标识别方法能够提高目标正确识别率,且其稳定性较好。该方法是一种有效的目标识别方法。Aiming at the problem of lower correct identification rate and stability of the support vector machine （SVM） identifying the high resolution range profile （HRRP）, the theory of target decomposition was studied and applied to decompose the radar target＇s polarization scatter matrix, and the target＇s structure characteristic profile was calculated. On the basic of analyzing and studying the methods of training and testing of SVM, the SVM was adopted to identify two experimental targets by their structure characteristic profile, and the identification result shown that the correct identification rate was improved by the target identification method of SVM based on the target structure characteristic profile and became much stabler. This is one kind of valid target identification method.

关 键 词：目标识别 结构特征像 支持向量机 高分辨一维距离像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]