Mutual-Information-Based Semi-Supervised Hyperspectral Band Selection With High Discrimination, High Information, and Low Redundancy

被引:115
作者
Feng, Jie [1 ]
Jiao, Licheng [1 ]
Liu, Fang [1 ]
Sun, Tao [1 ]
Zhang, Xiangrong [1 ]
机构
[1] Xidian Univ, Minist Educ China, Key Lab Intelligent Percept & Image Understanding, Xian 710071, Peoples R China
来源
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING | 2015年 / 53卷 / 05期
基金
中国国家自然科学基金; 新加坡国家研究基金会;
关键词
Clonal selection algorithm (CSA); hyperspectral band selection; multivariable mutual information (MMI); semi-supervised learning; DIMENSIONALITY REDUCTION; IMAGING SPECTROSCOPY; CLONAL SELECTION; REGISTRATION; IMAGES;
D O I
10.1109/TGRS.2014.2367022
中图分类号
P3 [地球物理学]; P59 [地球化学];
学科分类号
0708 ; 070902 ;
摘要
The large number of spectral bands in hyperspectral images provides abundant information to distinguish different land covers. However, these spectral bands have much redundancy and bring an extra computational burden. Thus, band selection is important for hyperspectral images. Since the labeled samples are difficult to obtain, a semi-supervised criterion based on maximum discrimination and information (MDI) is defined by using both limited labeled samples and sufficient unlabeled samples. This MDI criterion aims to select the most highly discriminative and informative bands, but it is hard to accurately calculate. Therefore, a novel criterion based on high discrimination, high information, and low redundancy (DIR) is proposed as its low-order approximation. Moreover, from an information theory perspective, a theoretical proof is given that many traditional semi-supervised feature selection criteria are the low-order approximations of this MDI criterion. Compared with them, the proposed criterion needs more relaxed approximation conditions. To search and optimize the proposed criterion, a novel clonal selection algorithm is proposed, where the adaptive clone and mutation operators are devised to speed up the convergence. Experimental results on hyperspectral images demonstrate the effectiveness of the proposed semi-supervised band selection method.
引用
收藏
页码:2956 / 2969
页数:14
相关论文
共 54 条
[1]  
Agarwal Abhishek, 2007, 2007 IEEE International Symposium on Signal Processing and Information Technology, P353, DOI 10.1109/ISSPIT.2007.4458191
[2]  
[Anonymous], 2006, 1530 U WISC MAD DEP
[3]  
[Anonymous], 2007, Hyperspectral Data Exploitation: Theory and Applications
[4]   Toward an optimal SVM classification system for hyperspectral remote sensing images [J].
Bazi, Yakoub ;
Melgani, Farid .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2006, 44 (11) :3374-3385
[5]   Multimodal Data Fusion Based on Mutual Information [J].
Bramon, Roger ;
Boada, Imma ;
Bardera, Anton ;
Rodriguez, Joaquim ;
Feixas, Miquel ;
Puig, Josep ;
Sbert, Mateu .
IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS, 2012, 18 (09) :1574-1587
[6]  
Brown A. J., 2006, THESIS MACQUARIE U S
[7]   Spectral curve fitting for automatic hyperspectral data analysis [J].
Brown, Adrian Jon .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2006, 44 (06) :1601-1608
[8]   Hyperspectral imaging spectroscopy of a Mars analogue environment at the North Pole Dome, Pilbara Craton, Western Australia [J].
Brown, AJ ;
Walter, MR ;
Cudahy, TJ .
AUSTRALIAN JOURNAL OF EARTH SCIENCES, 2005, 52 (03) :353-364
[9]   Graph Regularized Nonnegative Matrix Factorization for Data Representation [J].
Cai, Deng ;
He, Xiaofei ;
Han, Jiawei ;
Huang, Thomas S. .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2011, 33 (08) :1548-1560
[10]   Semi-supervised graph-based hyperspectral image classification [J].
Camps-Valls, Gustavo ;
Bandos, Tatyana V. ;
Zhou, Dengyong .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2007, 45 (10) :3044-3054