| 摘要: |
| 针对合成孔径雷达（Synthetic Aperture Radar，SAR）图像目标识别问题，提出结合多源特征和高斯过程模型的方法。分别利用主成分分析（Principal Component Analysis，PCA）、非负矩阵分解（Non-negative Matrix Factorization，NMF）以及单演信号提取SAR图像的特征矢量，并将它们串接为单一矢量。三类特征从不同角度描述SAR图像目标特性，从而为目标识别提供更为有效的信息。决策分类过程采用高斯过程模型进行多元分类，基于融合特征矢量获得概率意义上的最佳决策。实验中，采用MSTAR数据集设置3类目标、10类目标、型号差异以及俯仰角差异识别问题，结果验证了提出方法的优越性能。 |
| 关键词: 合成孔径雷达(SAR) 目标识别 主成分分析(PCA) 非负矩阵分解(NMF) 单演信号 高斯过程模型 |
| DOI： |
|
| 基金项目:广西创新驱动发展专项项目（桂科AA18242030） |
|
| Target recognition of SAR images by combining multiple features and Gaussian process model |
| XIN Haiyan,TONG Youwei |
| (School of Electronic Information and Automation,Guilin University of Aerospace Technology,Guilin 541004,China;School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China) |
| Abstract: |
| A target recognition method of synthetic aperture radar(SAR) image is developed via combination of multiple features and Gaussian process model.The principal component analysis(PCA),non-negative matrix factorization(NMF),and monogenic signal are employed for SAR image feature extraction separately and their results are fused as a single feature vector.The three features describe the target characteristics in SAR images from different aspects,thus providing more effective information for target recognition.In the decision stage,the Gaussian process model is used to perform the multi-class classification and the statically optimal decision can be made based on the fused feature vector.In the experiments,the MSTAR dataset is used to setup the 3-class,10-class,configuration variance,and depression angle variance problems.The results validate the superior performance of the proposed method. |
| Key words: synthetic aperture radar(SAR) target recognition principal component analysis(PCA) non-negative matrix factorization(NMF) monogenic signal Gaussian procedure process |
《电讯技术》编辑部 