基于多特征决策融合的SAR目标识别方法

Target recognition of SAR images via decision fusion of multiple features

文中提出基于多特征决策融合的合成孔径雷达(SAR)目标识别方法,不同特征对于描述原始SAR图像具有良好的互补性。文中选用主成分分析(PCA)、非负矩阵分解(NMF)以及核主成分分析(KPCA)对SAR图像进行特征提取。然后,采用稀疏表示分类器(SRC)对三类特征进行分类。最后,采用线性加权的形式对三者的决策变量进行融合得到稳健的识别结果。基于MSTAR数据库进行了目标识别实验,验证了方法的有效性。

This paper proposes a Synthetic Aperture Radar（ SAR） target recognition method via decision fusion of multiple features. Different features could provide complementary descriptions for the original SAR image. This study chooses Principal Component Analysis（ PCA）,Non-negative Matrix Factorization（ NMF）,and Kernel Principal Component Analysis（ KPCA） to extract features from SAR images.Afterwards,the Sparse Representation-based Classification（ SRC） is employed to classify the three features. Finally,a linear fusion strategy is used to fuse the decisions from the three features to make more robust decisions. Experiments are conducted on the MSTAR dataset to evaluate the validity of the proposed method.