Please wait a minute...
img

官方微信

遥感技术与应用  2020, Vol. 35 Issue (3): 634-644    DOI: 10.11873/j.issn.1004-0323.2020.3.0634
数据与图像处理     
基于Boosting的高光谱遥感切空间协同表示集成学习方法
虞瑶(),苏红军(),姚文静
河海大学地球科学与工程学院,江苏 南京 211100
Boosting Ensemble Learning for Hyperspectral Image Classification Using Tangent Collaborative Representation
Yao Yu(),Hongjun Su(),Wenjing Yao
School of Earth Sciences and Engineering, Hohai University, Nanjing 211100, China
 全文: PDF(6955 KB)   HTML
摘要:

近年来,协同表示分类(Collaborative Representation Classification,CRC)算法成为高光谱遥感影像分类的研究热点,尤其是切空间协同表示分类(Tangent Space Collaborative Representation,TCRC)利用切平面估计测试样本的局部流形,其分类精度得到了显著提高。为进一步提升高光谱遥感影像分类的准确性和可靠性,提出了基于Boosting的高光谱遥感影像切空间协同表示分类算法(Boosting-based Tangent Space Collaborative Representation Classification,Boost TCRC)。Boost TCRC算法采用TCRC算法作为基分类器,通过Boosting原理自适应地调整训练样本的权重,增大错分样本的权重从而使得分类器专注于较难分类的训练样本,然后在基于残差域融合时根据基分类器的分类表现赋予其权重,最终采用最小重构误差的原则对测试样本进行分类。实验采用HyMap(Hyperspectral Mapper)和AVIRIS(Airbone Visible Infrared Imaging Spectrometer)等高光谱遥感影像数据对所提出算法的性能进行了综合评价,结果表明:基于Boosting的集成方式可有效提升TCRC算法的分类效果。针对HyMap数据,Boost TCRC算法总体分类精度和Kappa系数分别为93.73%和0.920 8,两种精度指标分别高于TCRC算法2.82%和0.032 3,同时分别高于AdaBoost ELM算法1.81%和0.022 5。对于AVIRIS数据,Boost TCRC算法总体分类精度和kappa系数为84.11%和0.812 0,两种精度指标分别高于TCRC算法3.97%和0.049 3,同时分别高于AdaBoost ELM算法12.02%和0.143 6。

关键词: 切空间协同表示集成学习Boosting高光谱遥感分类    
Abstract:

Recently, Collaborative Representation Classification (CRC) has attracted much attention in hyperspectral image analysis. Due to uses the tangent plane to estimate the local manifold of the test sample. Tangent Collaborative Representation Classification (TCRC) achieve better performance. Furthermore, in order to improve the classification accuracy and reliability of hyperspectral remote sensing images, a novel Boosting-based Tangent Collaborative Representation ensemble method (Boost TCRC) for hyperspectral image classification is proposed. In this algorithm, Boost TCRC algorithm choose TCRC as base classifier and adjust the weight of the training samples adaptively by using the principle of Boosting. Increasing the weight of the misclassified samples so that the classifier concentrates on the training samples that are difficult to classify. Then assigns the weights according to the classification performance of the base classifier based on the residual domain fusion. Finally, the principle of minimum reconstruction error is adopted to classify the test sample. The performance of the proposed algorithm was comprehensively evaluated by hyperspectral remote sensing image data such as HyMap (Hyperspectral Mapper) and AVIRIS (Airbone Visible Infrared Imaging Spectrometer). The Boosting method can effectively improve the classification effect of the TCRC algorithm. For HyMap data, the overall classification accuracy and kappa coefficient of Boost TCRC algorithm are 93.73% and 0.920 8 respectively. Two precision values are higher than TCRC algorithm by 2.82% and 0.032 3, and are higher than the AdaBoost ELM algorithm by 1.81% and 0.022 5. For AVIRIS data, the overall classification accuracy and kappa coefficient of Boost TCRC algorithm are 84.11% and 0.8120 respectively. Two precision values are higher than TCRC algorithm by 3.97% and 0.049 3, and are higher than AdaBoost ELM algorithm by 12.02% and 0.143 6.

Key words: Tangent collaborative representation    Ensemble learning    Boosting    Hyperspectral image classification
收稿日期: 2019-04-01 出版日期: 2020-07-10
ZTFLH:  TP751  
基金资助: 国家自然科学基金项目“高光谱遥感影像多特征优化模型与协同表示分类”(41571325);“高光谱遥感表示模型与分类器动态集成方法”(41871220)
通讯作者: 苏红军     E-mail: yuyaoyao_yy@163.com;hjsu@hhu.edu.cn
作者简介: 虞瑶(1995-),女,安徽安庆人,硕士研究生,主要从事高光谱遥感影像分类研究。E?mail: yuyaoyao_yy@163.com
服务  
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章  
虞瑶
苏红军
姚文静

引用本文:

虞瑶,苏红军,姚文静. 基于Boosting的高光谱遥感切空间协同表示集成学习方法[J]. 遥感技术与应用, 2020, 35(3): 634-644.

Yao Yu,Hongjun Su,Wenjing Yao. Boosting Ensemble Learning for Hyperspectral Image Classification Using Tangent Collaborative Representation. Remote Sensing Technology and Application, 2020, 35(3): 634-644.

链接本文:

http://www.rsta.ac.cn/CN/10.11873/j.issn.1004-0323.2020.3.0634        http://www.rsta.ac.cn/CN/Y2020/V35/I3/634

图1  Purdue Campus 数据集
图2  Purdue Campus数据集6种分类算法的分类效果图
数据Purdue CampusIndian Pines
TCRCλ1e-61e-5
η1e-81e-8
n88
Boost TCRCλ1e-91e-9
η1e-81e-8
n88
T2020
表1  实验设置的最佳参数
图3  Indian Pines数据集
类别训练样本测试样本BoostingRFELMTCRCAdaBoost ELMBoost TCRC
道路151 27282.6082.2089.6788.9191.1392.44
草地151 09989.7890.9389.1285.5495.2989.49
阴影1520471.0394.0266.3190.4681.8788.43
土壤1536473.2494.4580.6686.8286.8591.19
树木151 33698.6893.4999.1598.1899.3398.89
建筑物151 27083.7279.6789.6094.6387.3296.17
总体分类精度/%86.0987.3189.3690.9191.9293.73
平均分类精度/%83.1889.1385.7590.7690.3092.77
Kappa系数0.825 70.841 30.866 50.888 50.898 30.920 8
时间/s1.250.100.1418.401.30126.95
表2  实验一分类精度统计
图4  Indian Pines数据集6种算法的分类效果图
类别训练样本测试样本BoostingRFELMTCRCAdaBoost ELMBoost TCRC
C1721 35655.4157.6962.8177.2564.7084.85
C24278862.2845.6554.2782.0353.7775.27
C32545873.5173.4487.2097.5990.8495.83
C43769389.2291.2394.6794.4796.9497.29
C52445494.0597.0599.0499.5299.80100
C64992358.4651.5356.8180.0155.6973.18
C7123233263.0679.6062.9066.0363.5475.16
C83056352.9135.9962.9381.4371.6784.19
C9641 20194.8195.2798.4598.4998.9399.14
总体分类精度/%69.4871.0571.5180.1472.0984.11
平均分类精度/%71.5369.7275.4586.3177.3287.21
Kappa系数0.637 10.655 30.662 10.762 80.668 40.812 0
时间/s12.210.350.13164.302.303 651.34
表3  实验二分类精度统计
图5  两组数据不同T值下算法的分类结果
图6  两组数据不同λ值下算法的分类结果
图7  两组数据不同η值下算法的分类结果
1 Tong Qingxi, Zhang Bing, Zheng Lanfen. Hyperspectral Remote Sensing [M]. Beijing: Higher Education Press,2006.
1 童庆禧,张兵,郑兰芬.高光谱遥感[M].北京:高等教育出版社,2006.
2 Landgrebe D. Hyperspectral Image Data Analysis[J]. IEEE Signal Processing Magazine, 2002, 19(1):17-28.
3 Moon H, Ahn H, Kodell R L, et al. Ensemble Methods for Classification of Patients for Personalized Medicine with High-dimensional Data[J]. Artificial Intelligence in Medicine, 2007, 41(3):197-207.
4 Bioucas-Dias J M, Plaza A, Camps-Valls G, et al. Hyperspectral Remote Sensing Data Analysis and Future Challenges[J]. IEEE Geoscience and Remote Sensing Magazine, 2013, 1(2):6-36.
5 Dozier J, Painter T H. Multispectral and Hyperspectral Remote Sensing of Alpine Snow Properties[J]. Annual Review Earth Planetary, Sciences, 2004, 32: 465-494.
6 Braun A C, Weidner U, Hinz S. Classification in High-dimensional Feature Spaces-assessment Using SVM, IVM and RVM with Focus on Simulated EnMAP Data[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2012, 5(2):436-443.
7 Chen C, Li W, Su H, et al. Spectral-spatial Classification of Hyperspectral Image based on Kernel Extreme Learning Machine[J]. Remote Sensing, 2014, 6(6): 5795-5814.doi: 10.3390/rs6065795.
doi: 10.3390/rs6065795
8 Zhang Y, Cao G, Li X, et al. Cascaded Random Forest for Hyperspectral Image Classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2018,11(4):1082-1094.
9 Li S, Song W, Fang L, et al. Deep Learning for Hyperspectral Image Classification:An Overview[J].IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(9):6690-6709.
10 Zhang H, Li Y, Jiang Y, et al. Hyperspectral Classification based on Lightweight 3-D-CNN With Transfer Learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(8): 5813-5828.
11 Chen Y, Wang Y, Gu Y, et al. Deep Learning Ensemble for Hyperspectral Image Classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(6): 1882-1897.
12 Bao R, Xia J, Mura M D, et al. Combining Morphological Attribute Profiles via an Ensemble Method for Hyperspectral Image Classification[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(3): 359-363.
13 Benediktsson J A, Chanussot J, Fauvel M. Multiple Classifier Systems in Remote Sensing: from Basics to Recent Developments[C]//International Workshop on Multiple Classifier Systems. Springer, Berlin, Heidelberg, 2007: 501-512.
14 Chan J C W, Paelinckx D. Evaluation of Random Forest and Adaboost Tree-based Ensemble Classification and Spectral Band Selection for Ecotope Mapping Using Airborne Hyperspectral Imagery[J]. Remote Sensing of Environment, 2008, 112(6): 2999-3011.doi: 10.1016/j.rse.2008.02.011.
doi: 10.1016/j.rse.2008.02.011
15 Gislason P O, Benediktsson J A, Sveinsson J R. Random Forests for Land Cover Classification[J]. Pattern Recognition Letters, 2006, 27(4): 294-300.
16 Xia J, Dalla Mura M, Chanussot J, et al. Random Subspace Ensembles for Hyperspectral Image Classification with Extended Morphological Attribute Profiles[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(9): 4768-4786.
17 Haq Q S U, Tao L, Yang S. Neural Network based Adaboosting Approach for Hyperspectral Data Classification[C]// IEEE International Conference on Computer Science & Network Technology, 2012.
18 Samat A, Du P, Liu S, et al. ELM2: Ensemble Extreme Learning Machines for Hyperspectral Image Classification[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(4): 1060-1069.
19 Chen Y, Zhao X, Lin Z. Joint Adaboost and Multifeature based Ensemble for Hyperspectral Image Classification[C]//2014 IEEE Geoscience and Remote Sensing Symposium, 2014: 2874-2877.
20 Xia J, Ghamisi P, Yokoya N, et al. Random Forest Ensembles and Extended Multiextinction Profiles for Hyperspectral Image Classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(1): 202-216.
21 Su H, Bo Z, Qian D, et al. Tangent Distance-based Collaborative Representation for Hyperspectral Image Classification[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(9):1236-1240.
22 Chi Y, Porikli F. Classification and Boosting with Multiple Collaborative Representations[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2014,36(8):1519-1531.
[1] 袁春琦,徐佳,程圆娥,陈媛媛,许康. 基于协同训练与集成学习的极化SAR图像半监督分类[J]. 遥感技术与应用, 2017, 32(2): 380-385.
[2] 鲍蕊,夏俊士,薛朝辉,杜培军,车美琴. 基于形态学属性剖面的高光谱影像集成分类[J]. 遥感技术与应用, 2016, 31(4): 731-738.