基于支持向量机和无源特征的目标识别方法  

Airplane Recognition Based on SVM and Passive Features

作　　者：史豪杰[1] 邢清华[1] 沈继承[1] 

机构地区：[1]空军工程大学导弹学院,陕西三原713800

出　　处：《电光与控制》2009年第2期36-38,80,共4页Electronics Optics & Control

摘　　要：通过引入机载雷达辐射这一无源特征,采用多类分类支持向量机进行类型识别,提出了一种更有效的目标识别方法。无源特征是有用信号和噪声的叠加,具有一定程度的不确定性,采用范数熵衡量无源特征,类间距较大,类内聚集性较强,还可以抑制噪声。支持向量机分类器结构简单、可获得全局最优、泛化能力强,多类分类支持向量机解决目标识别问题高效而且实用。实验证明,该方法明显地提高了目标识别的正确率。An effective method is put forward for airplane recognition by introducing in the passive feature of airborne radar radiation and using the multi-class Supports Vector Machine （SVM）for type identification. Passive feature is a mixture of signal and noise, which is uncertain. Norm entropy is used to measure the passive features, which has strong aggregation inside one type. It can also restrain noise. SVM has good characteristics of simple structure, global optimum and strong generalization ability, and Multi-SVM is effective and practical in airplane recognition. Experiment results showed that the method can improve the accuracy of airplane recognition.

关 键 词：目标识别 支持向量机 无源特征 范数熵 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计] TP18[自动化与计算机技术—控制理论与控制工程]