中文摘要：

针对当前高光谱影像稀疏分类模型中光谱重构方法单一性的问题,该文将稀疏分类模型中光谱的线性重构理解为光谱间的相似性度量,进而引入其他相似性测度指标,提出多相似测度稀疏表示的高光谱影像分类模型,并给出模型的统一解算方法——一般正交匹配追踪算法;随后,考虑地物空间连续性和一致性,将多相似测度稀疏分类模型扩展到空间联合的多相似测度稀疏分类模型,提出了一般联合匹配追踪算法;最后,利用两幅标准高光谱影像数据验证了所提出的多相似测度稀疏分类模型对于高光谱影像分类的有效性和实用性。

英文摘要：

In this paper, according to the spectral similarity of hyperspectral images and sparse representation, a novel framework of sparse representation classification model was proposed based on the general spectral similarity indices, such as spectral angle mapping （SAM） , spectral information divergence （SID） , spectral structure similarity index （SSIM） and so on. Then we presented a general algorithm called GOMP, inspired by the Orthogonal Matching Pursuit （OMP） algorithm. After that,considering the spatial features of the ground objects, we developed the GSRC to the joint GSRC model （GJSRC） and also extended the GSOMP algorithm to solve the proposed model. Finally,we demonstrated the practicality and effectiveness of the proposed algorithm through experiment of a real-world hyperspectral image data.