温兴平,胡光道,杨晓峰

• 1:中国地质大学数学地质遥感地质研究所
• 2:南京信息工程大学环境科学与工程学院

摘要(Abstract)：

CBERS卫星是由中国空间技术研究院与巴西空间研究院联合研制的地球资源遥感卫星,CBERS-02卫星数据总体质量比CBERS-01卫星有所提高,本文利用支持向量机方法对CBERS-02卫星影像信息进行提取。研究中首先用6S模式对影像进行大气校正,然后选择RBF为支持向量机方法的核函数,并用交叉验证方法得到影响RBF核函数的两个最佳参数值进行学习完成信息提取,最后将提取结果制作成矢量图。通过研究得出用大气校正后的数据进行信息提取分类精度有所提高;与最大似然法和最小距离法相比,支持向量机方法分类精度较高。通过将研究结果与ETM+影像进行比较得出,CBERS-02卫星影像精度能够满足应用需求并能代替TM/ETM+数据开展研究工作。

关键词(KeyWords)： CBERS-02卫星;支持向量机;信息提取

Abstract:

Keywords:

基金项目(Foundation): 国土资源大调查基金项目(编号:22003020002)

作者(Author): 温兴平,胡光道,杨晓峰

参考文献(References)：

• [1]邓乃扬,田英杰.数据挖掘中的新方法-支持向量机[M].北京:科学出版社,2004.
• [2]Vapnik V N.The Nature of Statistical Learning Theory(Second Edition)[M].NewYork:Springer,2000.
• [3]Cortes C,V Vapnik.Support-vector network[J].Ma-chine Learning,1995,20(3):273-297.
• [4]Scholkopf B,et al.Input Space Versus Feature Space inKernel-Based Methods[J].IEEE TRANSACTIONSON NEURAL NETWORKS,1999,10(5).
• [5]Chapelle O,P Haffner,V N Vapnik.Support vector ma-chines for histogram-based image classification[J].Neural Networks,IEEE Transactions on,1999,10(5).
• [6]Boser B E,I M Guyon,V N Vapnik.A training algo-rithm for optimal margin classifiers[C]//in Proceed-ings of the fifth annual workshop on Computational learn-ing theory,1992:144-152.
• [7]Cristianini N,J Shawe-Taylor,李国正,王猛,曾华军译.An Introduction to Support Vector Machines andOther Kernel-based Learning Methods(支持向量机导论)[M].北京:电子工业出版社,2004.
• [8]Foody G M,A Mathur.The use of small training setscontaining mixed pixels for accurate hard image classifi-cation:Training on mixed spectral responses for classifi-cation by a SVM[J].Remote Sensing of Environment,2006,103(2):179-189.
• [9]Nemmour H,Y Chibani.Multiple support vector ma-chines for land cover change detection:An application formapping urban extensions[J].ISPRS Journal of Photo-grammetry and Remote Sensing,2006,61(2).
• [10]Durbha,S S,R L King,N H Younan.Support vectormachines regression for retrieval of leaf area index frommultiangle imaging spectroradiometer[J].Remote Sens-ing of Environment,2007,107(1-2):348-361.
• [11]Zhu G,D G Blumberg.Classification using ASTER dataand SVM algorithms;:The case study of Beer Sheva,Israel[J].Remote Sensing of Environment,2002,80(2).
• [12]Mazzoni D,et al.A data-mining approach to associa-ting MISR smoke plume heights with MODIS fire meas-urements[J].Remote Sensing of Environment,2007,107(1-2)
• [13]Camps-Valls G,et al.Retrieval of oceanic chlorophyllconcentration with relevance vector machines[J].Re-mote Sensing of Environment,2006,105(1):23-33.
• [14]Su L,et al.Support vector machines for recognition ofsemi-arid vegetation types using MISR multi-angleimagery[J].Remote Sensing of Environment,2007,107(1-2):299-311.
• [15]Keuchel J,et al.Automatic land cover analysis for Te-nerife by supervised classification using remotely senseddata[J].Remote Sensing of Environment,2003,86(4).
• [16]郭建宁,等.CBERS-01/02卫星CCD图像相对辐射校正研究[J].中国科学:E辑,2005,35(B12).
• [17]Vermote E F,et al.Second Simulation of the SatelliteSignal in the Solar Spectrum,6S:An overview[J].IEEE Transactions on Geoscience and Remote Sensing,1997,35(3):675-686.
• [18]Vermote E F,et al.Second Simulation of the SatelliteSignal in the Solar Spectrum(6S),6S User Guide Ver-sion 2[EB/OL].1997:218.
• [19]易维宁,等.CBERS-02卫星CCD图像的大气订正[J].遥感学报,2006,10(5):703-708.
• [20]郭建宁,等.CBERS-02 CCD和SPOT-4 HRVIR1两个传感器的敦煌场地在轨绝对辐射定标及对比分析[J].遥感学报,2006,10(5):624-629.
• [21]Wu TF,C J Lin,R C Weng.Probability Estimates forMulti-class Classification by Pairwise Coupling[J].The Journal of Machine Learning Research,2004,(5).
• [22]Keerthi S S,C J Lin.Asymptotic Behaviors of SupportVector Machines with Gaussian Kernel[J].NeuralComputation,2003.,15(7):1667-1689.
• [23]Lin HT,C J Lin.A study on sigmoid kernels for SVMand the training of non-PSD kernels by SMO-typemethods.Technical report[R].2003,Department ofComputer Science,National Taiwan University.