Deep Learning Ensemble for Hyperspectral Image Classification

被引:112
作者
Chen, Yushi [1 ]
Wang, Ying [2 ]
Gu, Yanfeng [1 ]
He, Xin [1 ]
Ghamisi, Pedram [3 ]
Jia, Xiuping [4 ]
机构
[1] Harbin Inst Technol, Sch Elect & Informat Engn, Harbin 150001, Heilongjiang, Peoples R China
[2] Harbin Univ Sci & Technol, Higher Educ Key Lab Measure & Control Technol & I, Harbin 150080, Heilongjiang, Peoples R China
[3] Helmholtz Zentrum Dresden Rossendorf, Helmholtz Inst Freiberg Resource Technol, D-09599 Freiberg, Germany
[4] Univ New South Wales, Sch Engn & Informat Technol, Canberra, ACT 2600, Australia
关键词
Convolutional neural network (CNN); deep learning; ensemble; hyperspectral imagery classification; random subspace; SPECTRAL-SPATIAL CLASSIFICATION; NEURAL-NETWORKS; RANDOM FOREST; CLASSIFIERS; FRAMEWORK;
D O I
10.1109/JSTARS.2019.2915259
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
Deep learning models, especially deep convolutional neural networks (CNNs), have been intensively investigated for hyperspectral image (HSI) classification due to their powerful feature extraction ability. In the same manner, ensemble-based learning systems have demonstrated high potential to effectively perform supervised classification. In order to boost the performance of deep learning-based HSI classification, the idea of deep learning ensemble framework is proposed here, which is loosely based on the integration of deep learning model and random subspace-based ensemble learning. Specifically, two deep learning ensemble-based classification methods (i.e., CNN ensemble and deep residual network ensemble) are proposed. CNNs or deep residual networks are used as individual classifiers and random sub-spaces contribute to diversify the ensemble system in a simple yet effective manner. Moreover, to further improve the classification accuracy, transfer learning is investigated in this study to transfer the learnt weights from one individual classifier to another (i.e., CNNs). This mechanism speeds up the learning stage. Experimental results with widely used hyperspectral datasets indicate that the proposed deep learning ensemble system provides competitive results compared with state-of-the-art methods in terms of classification accuracy. The combination of deep learning and ensemble learning provides a significant potential for reliable HSI classification.
引用
收藏
页码:1882 / 1897
页数:16
相关论文
共 52 条
[1]   Classification of hyperspectral data from urban areas based on extended morphological profiles [J].
Benediktsson, JA ;
Palmason, JA ;
Sveinsson, JR .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2005, 43 (03) :480-491
[2]  
Bhattacharya U, 2003, PROC INT CONF DOC, P16
[3]   The impact of diversity on the accuracy of evidential classifier ensembles [J].
Bi, Yaxin .
INTERNATIONAL JOURNAL OF APPROXIMATE REASONING, 2012, 53 (04) :584-607
[4]   Random forests [J].
Breiman, L .
MACHINE LEARNING, 2001, 45 (01) :5-32
[5]   Hyperspectral Image Classification Using Dictionary-Based Sparse Representation [J].
Chen, Yi ;
Nasrabadi, Nasser M. ;
Tran, Trac D. .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2011, 49 (10) :3973-3985
[6]   Hyperspectral Images Classification With Gabor Filtering and Convolutional Neural Network [J].
Chen, Yushi ;
Zhu, Lin ;
Ghamisi, Pedram ;
Jia, Xiuping ;
Li, Guoyu ;
Tang, Liang .
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 2017, 14 (12) :2355-2359
[7]   Deep Feature Extraction and Classification of Hyperspectral Images Based on Convolutional Neural Networks [J].
Chen, Yushi ;
Jiang, Hanlu ;
Li, Chunyang ;
Jia, Xiuping ;
Ghamisi, Pedram .
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 2016, 54 (10) :6232-6251
[8]   Deep Learning-Based Classification of Hyperspectral Data [J].
Chen, Yushi ;
Lin, Zhouhan ;
Zhao, Xing ;
Wang, Gang ;
Gu, Yanfeng .
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2014, 7 (06) :2094-2107
[9]   Optimizing Subspace SVM Ensemble for Hyperspectral Imagery Classification [J].
Chen, Yushi ;
Zhao, Xing ;
Lin, Zhouhan .
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 2014, 7 (04) :1295-1305
[10]  
Deng L, 2013, IEEE INT NEW CIRC