支持向量机的全局局部特征融合目标识别

Target Recognition Based on Support Vector Machine (SVM) Features Fusion

提出一种基于支持向量机的全局局部特征融合目标识别方法,并将其运用到雷达一维距离像目标识别。该方法采用非线性辨别方法与局部保留映射方法分别提取样本的非线性全局特征与局部特征,并进行特征融合,以便提取更全面的样本特征,得到更加准确的识别结果,随后采用支持向量机进行分类识别,利用其对于非线性小样本问题的强大处理能力,进一步改善识别结果。对三种飞机目标的实测雷达一维距离像进行了仿真实验,结果表明了方法的有效性。

This paper proposes a target recognition method based on support vector machine features fusion. The method uses nonlinear discrimination analysis and local retain mapping to extract the global and local features and then makes features fusion in order to extract more comprehensive samples and obtain more accurate identification results. Then the support vector machine is used for classification. Since its power to deal with nonlinear and small samples, the identification accuracy is further improved. The simulation results of three plane targets show the effectiveness.