基于mRMR特征优选算法的多光谱遥感影像分类效率精度分析

doi:10.3724/SP.J.1047.2016.00815

地球信息科学学报 ›› 2016, Vol. 18 ›› Issue (6): 815-823.doi: 10.3724/SP.J.1047.2016.00815

基于mRMR特征优选算法的多光谱遥感影像分类效率精度分析

程希萌^1,²(), 沈占锋^2,^*(), 邢廷炎¹, 夏列钢³, 吴田军⁴

1. 中国地质大学（北京）信息工程学院,北京 100083
2. 中国科学院遥感与数字地球研究所,北京 100101
3. 浙江工业大学计算机科学与技术学院,杭州 310023
4. 长安大学理学院,西安 710064

收稿日期:2015-10-27 修回日期:2015-12-04 出版日期:2016-06-10 发布日期:2016-06-10
通讯作者: 沈占锋 E-mail:cheng_ximeng@126.com;shenzf@radi.ac.cn
作者简介:
作者简介：程希萌（1991-）,男,硕士生,研究方向为空间数据挖掘。E-mail: cheng_ximeng@126.com
基金资助:
国家高分辨率对地观测系统重大专项(03-Y30B06-9001-13/15-01);中国科学院重点部署项目(KZZD-EW-07-02);国家高技术研究发展项目(2013AA12A401)

Efficiency and Accuracy Analysis of Multispectral Image Classification Based on mRMR Feature Selection Method

CHENG Ximeng^1,²(), SHEN Zhanfeng^2,^*(), XING Tingyan¹, XIA Liegang³, WU Tianjun⁴

1. School of Information Engineering, China University of Geosciences, Beijing 100083, China
2. Institute of Remote Sensing and Digital Earth, CAS, Beijing 100101, China
3. College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou 310023, China
4. College of Science, Chang’an University, Xi’an 710064, China

Received:2015-10-27 Revised:2015-12-04 Online:2016-06-10 Published:2016-06-10
Contact: SHEN Zhanfeng E-mail:cheng_ximeng@126.com;shenzf@radi.ac.cn

摘要/Abstract

摘要：

在遥感图像分类过程中,进行合理的特征优选操作,将有助于提高分类器的分类效率及精度。本文以淮南地区资源三号卫星多光谱遥感影像数据为例,采用二值离散化、直方图法及F统计法3种计算方法实现mRMR（minimal-Redundancy-Maximal-Relevance）算法特征优选过程。根据3种方法所得到的特征优选结果及全部特征信息,分别采用C5.0决策树和K近邻2种分类器进行图像分类实验,并利用目视解译方法对不同方法组合的影像分类结果进行精度验证。实验结果表明,利用3种计算方法实现mRMR特征优选算法对不同分类器的影响程度不同：在分类效率方面,C5.0决策树分类器可提高36.84%,而K近邻分类器可提高72.05%;在分类精度方面,C5.0决策树分类器能保证分类精度大致不变,总体分类精度可提高0.60%,Kappa系数可提高0.80%,而K近邻分类器总体分类精度可提高4.34%,Kappa系数可提高7.90%。

关键词: mRMR算法, 多光谱影像, 互信息, 特征优选, 图像分类

Abstract:

Image classification is a popular research topic in the field of remote sensing. This technology has been widely used in environmental protection, military, urban planning, and other fields. Interfering by the massive feature information of remote sensing image, applying the reasonable feature selection approach in the progress of image classification becomes critical for improving the efficiency and accuracy of classification. This paper extracts the image feature data from the ZY3 satellite multispectral image of Huainan region, and studies the mRMR (minimal-Redundancy-Maximal-Relevance) feature selection method. This algorithm has a simple core principle and low requirement of data. The core problem of this algorithm is the computation of mutual information. The mRMR algorithm is initially applied in the field of bioscience, such as the gene expression analysis, and it is not widely used in the field of remote sensing. This research uses three methods (the binary discretization, histogram method and F-statistic) to realize the computation process of mRMR algorithm. And two classifiers (the C5.0 decision tree and k-nearest neighbour) are used for the classification based on three types of feature selection results and the total feature information. Moreover, the visual interpretation is used to verify the image classification results from these different methods. The study shows that the results produced by different mRMR computation processes are distinct regarding to different classifiers. In terms of efficiency, all methods can improve the efficiency of C5.0 and KNN. The classification efficiency is increased by 36.84% for C5.0 and by 72.05% for KNN. In terms of accuracy, all method can maintain the accuracy of C5.0 while improve the accuracy of KNN. The total classification accuracy and Kappa coefficient are increased for C5.0 by 0.60% and 0.80%, respectively. The total classification accuracy is increased by 4.34% and the Kappa coefficient is increased by 7.90% for KNN. In summary, the feature selection method based on the mRMR algorithm is effective in the procedure of multispectral image classification.

Key words: mRMR algorithm, multispectral image, mutual information, feature selection, image classification

程希萌, 沈占锋, 邢廷炎, 夏列钢, 吴田军. 基于mRMR特征优选算法的多光谱遥感影像分类效率精度分析[J]. 地球信息科学学报, 2016, 18(6): 815-823.DOI:10.3724/SP.J.1047.2016.00815

CHENG Ximeng,SHEN Zhanfeng,XING Tingyan,XIA Liegang,WU Tianjun. Efficiency and Accuracy Analysis of Multispectral Image Classification Based on mRMR Feature Selection Method[J]. Journal of Geo-information Science, 2016, 18(6): 815-823.DOI:10.3724/SP.J.1047.2016.00815

图/表 10

图1

图2

图3

表1

表2

图4

图5

表3

表4

表5

参考文献 24

[1]	李德仁. 论21世纪遥感与GIS的发展[J].武汉大学学报·信息科学版,2003,28(2):127-131.
	[ Li D R.Towards the development of remote sensing and GIS in the 21st century[J]. Geomatics and Information Science of Wuhan University, 2003,28(2):127-131. ]
[2]	田静,王卷乐,李一凡,等.基于决策树方法的蒙古高原土地覆盖遥感分类——以蒙古国中央省为例[J].地球信息科学学报,2014,16(3):460-469. doi: 10.3724/SP.J.1047.2014.00460
	[ Tian J, Wang J L, Li Y F, et al.Land cover classification in Mongolian Plateau based on decision tree method: a case study in Tov Province, Mongolia[J]. Journal of Geo-information Science, 2014,16(3):460-469. ] doi: 10.3724/SP.J.1047.2014.00460
[3]	许夙晖,慕晓冬,柯冰,等.基于遥感影像的军事阵地动态监测技术研究[J].遥感技术与应用,2014,29(3):511-516. doi: 10.11873/j.issn.1004-0323.2014.3.0511
	[ Xu S H, Mu X D, Ke B, et al.Dynamic monitoring of military position based on remote sensing image[J]. Remote Sensing Technology and Application, 2014,29(3):511-516. ] doi: 10.11873/j.issn.1004-0323.2014.3.0511
[4]	李淑娟,王黎明,董南.城市建筑物人口时空分布模型与实验分析——以北京东华门街道为例[J].地球信息科学学报,2013,15(1):19-28. doi: 10.3724/SP.J.1047.2013.00019
	[ Li S J, Wang L M, Dong N.Simulation of urban small-area population space-time distribution based on building extraction: taking Beijing Donghuamen subdistrict as an example[J]. Journal of Geo-information Science, 2013,15(1):19-28. ] doi: 10.3724/SP.J.1047.2013.00019
[5]	Sotoca J M, Pla F, Sánchez J S.Band selection in multispectral images by minimization of dependent information[J]. IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews, 2007,37(2):258-267.
[6]	王露,龚光红.基于ReliefF+mRMR特征降维算法的多特征遥感图像分类[J].中国体视学与图像分析,2014,19(3):250-257.
	[ Wang L, Gong G H.Multiple features remote sensing image classification based on combining ReliefF and mRMR[J]. Chinese Journal of Stereology and Image Analysis, 2014,19(3):250-257. ]
[7]	张鹏. 基于主成分分析的综合评价研究[D].南京:南京理工大学,2004.
	[ Zhang P.Comprehensive assessment study based on principal component analysis[D]. Nanjing: Nanjing University of Science and Technology, 2004. ]
[8]	史振威. 独立成分分析的若干算法及其应用研究[D].大连:大连理工大学,2005.
	[ Shi Z W.Several algorithms for independent component analysis and their applications[D]. Dalian: Dalian University of Technology, 2005. ]
[9]	李波. 基于流形学习的特征提取方法及其应用研究[D].合肥:中国科学技术大学,2008.
	[ Li B.The study of the manifold learning based feature extraction methods and their applications[D]. Hefei: University of Science and Technology of China, 2008. ]
[10]	Peng H, Long F, Ding C.Feature selection based on mutual information: criteria of max-dependency, max-relevance, and min-redundancy[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005,27(8):1226-1238.
[11]	姚旭,王晓丹,张玉玺,等.特征选择方法综述[J].控制与决策,2012,27(2):161-166, 192.
	[ Yao X, Wang X D, Zhang Y X, et al.Summary of feature selection algorithms[J]. Control and Decision, 2012,27(2):161-166,192. ]
[12]	Ding C, Peng H.Minimum redundancy feature selection from microarray gene expression data[J]. Journal of Bioinformatics and Computational Biology, 2005,3(2):185-205. doi: 10.1109/CSB.2003.1227396 pmid: 15852500
[13]	吴波,朱勤东,高海燕,等.面向对象影像分类中基于最大化互信息的特征选择[J].国土资源遥感, 2009,21(3):30-34. doi: 10.6046/gtzyyg.2009.03.06
	[ Wu B, Zhu Q D, Gao H Y, et al.Feature selection based on maxmal mutual information criterion in object-oriented classification[J]. Remote Sensing for Land & Resources, 2009,21(3):30-34. ] doi: 10.6046/gtzyyg.2009.03.06
[14]	邹利东,潘耀忠,朱文泉,等.结合邻域相关影像与最大相关性最小冗余性特征选择的面向对象变化检测[J].中国图象图形学报,2014,19(1):158-166. doi: 10.11834/jig.20140120
	[ Zou L D, Pan Y Z, Zhu W Q, et al.The object-oriented change detection based on neighborhood correlation images and the minimum-redundancy-maximum-relevance feature selection[J]. Journal of Image and Graphics, 2014,19(1):158-166. ] doi: 10.11834/jig.20140120
[15]	李亦农,李梅.信息论基础教程[M].北京:北京邮电大学出版社,2005.
	[ Li Y N, Li M.Information theory fundamentals[M]. Beijing: Beijing University of Posts and Telecommunications Press, 2005. ]
[16]	龚伟. 基于信息熵和互信息的流域水文模型不确定性分析[D].北京:清华大学,2012.
	[ Gong W.Watershed model uncertainty analysis based on information entropy and mutual information[D]. Beijing: Tsinghua University, 2012. ]
[17]	Dudoit S, Fridlyand J, Speed T P.Comparison of discrimination methods for the classification of tumors using gene expression data[J]. Journal of the American Statistical Association, 2002,97(457):77-87. doi: 10.1198/016214502753479248
[18]	Wu X, Kumar V, Quinlan J R, et al.Top 10 algorithms in data mining[J]. Knowledge and Information Systems, 2008,14(1):1-37. doi: 10.1007/s10115-007-0114-2
[19]	Harrington P.机器学习实战[M].北京:人民邮电出版社,2013.
	[ Harrington P.Machine learning in action[M]. Beijing: Posts & Telecom Press, 2013. ]
[20]	卢东标. 基于决策树的数据挖掘算法研究与应用[D].武汉:武汉理工大学,2008.
	[ Lu D B.Research and application on the data mining algorithm based on decision tree[D]. Wuhan: Wuhan University of Technology, 2008. ]
[21]	杨显明,焦华富,许吉黎.基于发生学视角的淮南城市空间生长过程、特征及影响因素研究[J].地理科学,2014,34(5):563-570.
	[ Yang X M, Jiao H F, Xu J L.Process, characteristics and influence factors of the Huainan City’s spatial expansion from the embryology perspective[J]. Scientia Geographica Sinica, 2014,34(5):563-570. ]
[22]	孙贤斌. 淮南市土壤重金属污染生态研究[D].芜湖:安徽师范大学,2003.
	[ Sun X B.Ecological study on the soil heavy metal pollution in Huainan city[D]. Wuhu: Anhui Normal University, 2003. ]
[23]	吴田军,骆剑承,夏列钢,等.迁移学习支持下的遥感影像对象级分类样本自动选择方法[J].测绘学报,2014,43(9):908-916. doi: 10.13485/j.cnki.11-2089.2014.0163
	[ Wu T J, Luo J C, Xia L G, et al.An automatic sample collection method for object-oriented classification of remotely sensed imageries based on transfer learning[J]. Acta Geodaetica et Cartographica Sinica, 2014,43(9):908-916. ] doi: 10.13485/j.cnki.11-2089.2014.0163
[24]	Kwak N, Choi C H.Input feature selection by mutual information based on Parzen window[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24(12):1667-1671. doi: 10.1109/TPAMI.2002.1114861

特征类型	特征名称	特征数目
光谱特征	第1-4波段均值、第1-4波段标准差、NDWI均值、NDWI标准差、NDVI均值、NDVI标准差	12
几何特征	长、宽、长宽比、像元数目、边长、形状指数、角点数目	7
空间关系	紧致度、主方向	2
纹理特征	同质性、对比度、相异性、角二阶矩、熵、相关性	6

计算方法	优选特征名称	优选特征数目
BD	NDVI均值、第1波段均值、NDWI均值、第3波段均值、第4波段均值、第1波段标准差、第2波段均值、角二阶矩、相关性、同质性、第2波段标准差、第3波段标准差、相异性、对比度、熵、第4波段标准差、NDWI标准差、NDVI标准差、长宽比、主方向	20
HM	NDWI均值、第1波段均值、相关性、同质性、第4波段均值、第1波段标准差、NDVI均值、第3波段均值、第4波段标准差、角二阶矩、第2波段均值、相异性、第3波段标准差、第2波段标准差、熵、对比度、NDVI标准差、NDWI标准差	18
F	NDWI均值、NDVI均值、第4波段均值、第3波段均值、相关性	5

方法组合	分类时间/s	总体分类精度/(%)	Kappa系数
C5.0	1.349	83.5	0.747
C5.0+BD	1.209	82.0	0.724
C5.0+HM	1.157	84.0	0.753
C5.0+F	0.852	82.0	0.725

方法组合	分类时间/s	总分类精度/(%)	Kappa系数
KNN	276.014	80.5	0.696
KNN+BD	212.306	84.0	0.751
KNN+HM	195.178	82.0	0.719
KNN+F	77.156	81.0	0.705

方法组合	分类时间/s	总体分类精度/(%)	Kappa系数
C5.0	1.349	83.5	0.747
C5.0+mRMR	1.073	82.7	0.734
C5.0+PCA	0.924	78.0	0.658
KNN	276.014	80.5	0.696
KNN+mRMR	161.547	82.3	0.725
KNN+PCA	88.303	80.0	0.689

基于mRMR特征优选算法的多光谱遥感影像分类效率精度分析

Efficiency and Accuracy Analysis of Multispectral Image Classification Based on mRMR Feature Selection Method

RichHTML

PDF (PC)

赞

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 24

相关文章 6

Metrics

本文评价

推荐阅读 0

[1]	李玉, 李奕燃, 王光辉, 石雪. 基于加权指数函数模型的高光谱图像分类方法[J]. 地球信息科学学报, 2020, 22(8): 1642-1653.
[2]	王学文, 赵庆展, 韩峰, 马永建, 龙翔, 江萍. 机载多光谱影像语义分割模型在农田防护林提取中的应用[J]. 地球信息科学学报, 2020, 22(8): 1702-1713.
[3]	何昭欣, 张淼, 吴炳方, 邢强. Google Earth Engine支持下的江苏省夏收作物遥感提取[J]. 地球信息科学学报, 2019, 21(5): 752-766.
[4]	张沁雨,李哲,夏朝宗,陈健,彭道黎. 高分六号遥感卫星新增波段下的树种分类精度分析[J]. 地球信息科学学报, 2019, 21(10): 1619-1628.
[5]	梁艳平, 毛政元, 邹为彬, 许锐. 基于相似数据聚合与变K值KNN的短时交通流量预测[J]. 地球信息科学学报, 2018, 20(10): 1403-1411.
[6]	李雪轲, 王晋年, 张立福, 杨杭, 刘凯. 面向对象的航空高光谱图像混合分类方法[J]. 地球信息科学学报, 2014, 16(6): 941-948.