多分辨率遥感图像复合分类方法研究
|
摘要
本论文针对单一分辨率遥感图像分类中所普遍存在的低分辨率遥感图像混合像元解析能力差、分类精度低，高分辨率遥感图像分类处理时间长、同时相广覆盖高分辨率数据获取困难等实际问题，就多分辨率遥感图像的复合分类技术展开了深入研究。主要研究成果与创新点如下：
1、提出了一种新的多分辨率遥感图像复合分类总体框架，并给出了其基本处理流程。该框架通过广覆盖低分辨率遥感图像与同区域中少量高分辨率图像的有机结合，实现了高分辨率分类精度级别的、广域全局分类。
2、提出了一种基于非线性退化模型的复合分类方法。该方法采用非线性函数描述多分辨率遥感图像间的空间分辨率退化关系，并将组合核函数的概念用于非线性退化函数的参数求取，较好解决了非线性函数模型参数难以采用传统回归分析求取的难点。
3、提出了一种基于“真实”似然特征的复合分类算法。该算法在“真实”似然分布信息的提取过程中，利用灰度云理论，通过加权平滑窗算法提取“真实”似然特征，并提出了似然特征向量的加权最大似然分类准则，实现了空间分辨率退化关系的非参数化描述。
4、提出了一种基于条件随机场模型的复合分类算法。通过用来描述光谱特征与类别关系的真实似然特征序列生成模型、以及用来描述地物空间连续性的上下文关系模型，组成条件随机场模型的两类势函数，实现了子像元、区域级的高精度复合分类。
5、面向实际生产应用需求，设计实现了多分辨率遥感图像复合分类软件系统，构建了可用于业务运行的、基于复合分类的农作物种植面积估计生产模型与流程，并通过水稻种植面积估计实验，验证分析了本论文方法的实际应用精度。
In recent years, the integration and composited analysis of multi-resource satellitedata has become one of the most important techniques in the remote sensing field. Forlarge scale land covering classification, it’s no surprise that low resolution data hasworse performance due to the mixed-pixel problem, and high resolution data with widecovering range has more limitations such as long period of acquiring cycle, high dataand processing cost. Facing the above problems, a new composited classificationframework for the multi-resolution satellite data with in the same area and differentcovering ranges has been studied in this dissertation. To study the multi-to-single spatialcorrespondence between high resolution image and low resolution image,nonparametric “real” likelihood distribution estimation is adopted and “real” likelihoodfeatures for low resolution pixels are extracted based on cloud theory. An sub-pixellevel enhanced composited classification method based on multi-kernel non-linearregression model and a context level enhanced composited classification method basedon Conditional Random Fields model are proposed. Our experiments on different sets ofsatellite data show that the proposed methods can greatly improve the accuracy for largescale land covering classification applications. The major contributions of thisdissertation are listed as following:
A novel integrated frame of the composited classification method at algorithmlevel is proposed and realized. To solve the problems of tradition compositedclassification method based on linear regression model, an enhanced compositedclassification method based on multi-kernel non-linear regression model is proposed.
Several key techniques such as multi-to-single spatial relation between the higherand lower resolution satellite data, generation and description for the likelihooddistribution scatter plot and real-likelihood function extraction algorithms are putforward. And the relationships between real-likelihood function models andclassification accuracy, the relationships between the extending principles andclassification accuracy are analyzed in details.
A composited classification model which integrating spatial contextual informationbetween pixels and multi-to-single spatial correspondence is proposed. The sequences of “real” likelihood features, which represent relations between spectrum and landcovering types, is integrated into the classifier with the spatial contextual informationbetween pixels by defining two types of potential functions. CRFs based classifieroffers a robust and accurate framework which can support multiple features andrepresents the special continuity of land covering.
Using various data combination and different kinds of ground truth data, theaccuracy of the proposed compounded classification algorithms has been analyzed, andthe results show that the proposed method can greatly improve the accuracy for largescale land covering classification applications.
The related functional modules and simulation software are developed and tested.An operational process for the Crop Condition Monitoring System is developed.
These researches have provided new efficient analysis methods for multi-resourcesatellite data and will extend the applications of composited classification to someextent.
1、提出了一种新的多分辨率遥感图像复合分类总体框架，并给出了其基本处理流程。该框架通过广覆盖低分辨率遥感图像与同区域中少量高分辨率图像的有机结合，实现了高分辨率分类精度级别的、广域全局分类。
2、提出了一种基于非线性退化模型的复合分类方法。该方法采用非线性函数描述多分辨率遥感图像间的空间分辨率退化关系，并将组合核函数的概念用于非线性退化函数的参数求取，较好解决了非线性函数模型参数难以采用传统回归分析求取的难点。
3、提出了一种基于“真实”似然特征的复合分类算法。该算法在“真实”似然分布信息的提取过程中，利用灰度云理论，通过加权平滑窗算法提取“真实”似然特征，并提出了似然特征向量的加权最大似然分类准则，实现了空间分辨率退化关系的非参数化描述。
4、提出了一种基于条件随机场模型的复合分类算法。通过用来描述光谱特征与类别关系的真实似然特征序列生成模型、以及用来描述地物空间连续性的上下文关系模型，组成条件随机场模型的两类势函数，实现了子像元、区域级的高精度复合分类。
5、面向实际生产应用需求，设计实现了多分辨率遥感图像复合分类软件系统，构建了可用于业务运行的、基于复合分类的农作物种植面积估计生产模型与流程，并通过水稻种植面积估计实验，验证分析了本论文方法的实际应用精度。
In recent years, the integration and composited analysis of multi-resource satellitedata has become one of the most important techniques in the remote sensing field. Forlarge scale land covering classification, it’s no surprise that low resolution data hasworse performance due to the mixed-pixel problem, and high resolution data with widecovering range has more limitations such as long period of acquiring cycle, high dataand processing cost. Facing the above problems, a new composited classificationframework for the multi-resolution satellite data with in the same area and differentcovering ranges has been studied in this dissertation. To study the multi-to-single spatialcorrespondence between high resolution image and low resolution image,nonparametric “real” likelihood distribution estimation is adopted and “real” likelihoodfeatures for low resolution pixels are extracted based on cloud theory. An sub-pixellevel enhanced composited classification method based on multi-kernel non-linearregression model and a context level enhanced composited classification method basedon Conditional Random Fields model are proposed. Our experiments on different sets ofsatellite data show that the proposed methods can greatly improve the accuracy for largescale land covering classification applications. The major contributions of thisdissertation are listed as following:
A novel integrated frame of the composited classification method at algorithmlevel is proposed and realized. To solve the problems of tradition compositedclassification method based on linear regression model, an enhanced compositedclassification method based on multi-kernel non-linear regression model is proposed.
Several key techniques such as multi-to-single spatial relation between the higherand lower resolution satellite data, generation and description for the likelihooddistribution scatter plot and real-likelihood function extraction algorithms are putforward. And the relationships between real-likelihood function models andclassification accuracy, the relationships between the extending principles andclassification accuracy are analyzed in details.
A composited classification model which integrating spatial contextual informationbetween pixels and multi-to-single spatial correspondence is proposed. The sequences of “real” likelihood features, which represent relations between spectrum and landcovering types, is integrated into the classifier with the spatial contextual informationbetween pixels by defining two types of potential functions. CRFs based classifieroffers a robust and accurate framework which can support multiple features andrepresents the special continuity of land covering.
Using various data combination and different kinds of ground truth data, theaccuracy of the proposed compounded classification algorithms has been analyzed, andthe results show that the proposed method can greatly improve the accuracy for largescale land covering classification applications.
The related functional modules and simulation software are developed and tested.An operational process for the Crop Condition Monitoring System is developed.
These researches have provided new efficient analysis methods for multi-resourcesatellite data and will extend the applications of composited classification to someextent.
引文
[1] Lurie J, Irvin E M. Remote sensing technology and applications. OpticalEngineering,2002,41(9):2075-2076.
[2] Gat N, Subramanian S, Barhen J, et al. Spectral Imaging Applications: RemoteSensing, Environmental Monitoring, Medicine, Military Operations, FactoryAutomation and Manufacturing. Proceeding of SPIE on Emerging Applications ofComputer Vision, Volume2962. The International Society for Optical Engineering,1997:63-77.
[3]朱述龙,朱宝山,王红卫.遥感图像处理与应用.北京:科学出版社,2006.
[4] Bauer, M E. Technological Basis and Applications of Remote Sensing of the Earth'sResources. IEEE Transactions on Geoscience Electronics,1976,14(1):3-9.
[5] Du P J, Zhang H P, Pan C, et al. Applications of Multi-source Remote SensingInformation to Urban Environment Monitoring in Mining Industrial Cities. UrbanRemote Sensing Joint Event,2007:1-12.
[6]钱永兰.运行化作物遥感识别中的多源数据组织与融合方法研究[D].北京:中国科学院,2005:1-13.
[7]孙卫东.基于多分辨率遥感图像的复合分类方法研究[国家自然科学基金结题报告(课题编号:60472029)].北京:清华大学电子系,2008.
[8]熊育久,林辉,孙华,等.多源遥感数据融合及其对植被识别的影响.林业资源管理,2005(5):71-76.
[9]李军,周月琴,李德仁.小波变换用于高分辨率全色影像与多光谱影像的融合研究.遥感学报,1999,3(2):116-120.
[10]陈思锦,秦其明,王文君.利用小波分析改进Brovey遥感影像融合方法.测绘学院学报,2004,21(2):118-120.
[11]董毓敏.基于小波叠加的主成分变换遥感数据融合方法的研究.东北测绘,2002,25(3):10-11.
[12]张炳智,张继贤,张丽.土地利用动态遥感监测中多源遥感影像融合方法比较研究.测绘科学,2000(3):46-51.
[13]贾永红,孙家柄.遥感多光谱影像数据与航片数字化影像融合方法的研究.测绘通报,1997(5):10-12.
[14]金剑秋,王章野,江照意,等.多光谱图像的真实感融合.中国图像图形学报,2002,7(A)(9):926-931.
[15]王智均,李德仁,李清泉.利用小波变换对影像进行融合的研究.武汉测绘科技大学学报,2000,25(2):137-142.
[16]张炳智,张继贤,张丽.土地利用动态遥感监测中多源遥感影像融合方法比较研究.测绘科学,2000(3):46-51.
[17] Tax T, Solberg S. Information Fusion in Remote Sensing. Vista in Astronomy,1997,41(3):337-342.
[18] Ronald Y. A framework for multi-source data fusion. Information Sciences,2004,163:175-200.
[19] Landgrebe D. On the relationship between class definition precision andclassification accuracy in hyperspectral analysis. IGARSS,2000:147-149.
[20] Harsanyi J, Chang C. Hyperspectral image classification and dimensionalityreduction：an orthogonal subspace projection approach. IEEE Trans. on G. R. S.1994,32(4):779-785.
[21]蔡学良,崔远来.基于异源多时相遥感数据提取灌区作物种植结构.北京:农业工程学报,2009,25(8):124-131.
[22] Jia X, Richards J. Segmented principal components transformation for efficienthyperspectral remote sensing image display and classification. IEEE Trans. On G. R.S.1999,37(1):538-542.
[23]张钧萍,张晔.基于多特征多分辨率融合的高光谱图像分类.红外与毫米波学报.2004,23(5):345-348.
[24] Zhang Y, Desai M, Zhang J, et al. Adaptive subspace decomposition forhyperspectral data dimensionality reduction. ICIP.1999:326-329.
[25]黄宁，朱敏慧，张守融，一种采用高斯隐马尔可夫随机场模型的遥感图像分类算法，电子与信息学报,2003,25(1):50-53.
[26]郭琳.基于多分辨率遥感数据的作物复合分类方法.北京:清华大学硕士学位论文,2008.
[27] DeFries R, Hansen M, Steininger M, Dubayah R, Sohlberg R and Townshend J.Subpixel forest cover in central Africa from multisensor, multitemporal data.Remote Sensing of Environment.1997,60:228–246.
[28]强强.不同分辨率遥感图像在地表分类中的综合应用.北京:清华大学电子系毕业设计论文,2003.
[29] Zhukov B, Oertel D, and Lanzl F. A multiresolution multisensor technique forsatellite remote sensing. IGARSS,1995,51-53.
[30] Cihlar J, Latifovic R, Chen J, Beaubien J and Li Z. Selecting Representative HighResolution Sample Images for Land Cover Studies. Part2: Application to EstimatingLand Cover Composition. Remote Sensing of Environment.2000,72:127-138.
[31] Rokhmatuloh, Bilbisi H A, Arihara K, et al. Application of Regression Tree Methodfor Estimating Percent Tree Cover of Asia with QuickBird images as training data.Proceedings of the CEReS International Symposium on Remote Sensing.2005.
[32]钱永兰.运行化作物遥感识别中的多源数据组织与融合方法研究[D].北京:中国科学院硕士学位论文,2005:1-13.
[33] Cushnie J L. The interactive effect of spatial resolution and degree of internalvariability within land-cover types on classification accuracies. International Journalof Remote Sensing,1987,8(1):15-29.
[34] Chen D M A multi-resolution analysis and classification framework for improvingland use/cover mapping from earth observation data. The International Archives ofthe Photogrammetry, Remote Sensing and Spatial Information Sciences,2004,34.
[35] Latty R S and Hoffer R M. Computer based classification accuracy due to the spatialresolution using per-point versus per-field classification techniques. Proc. ofRemotely Sensed Data Symposium, West Lafayette, Indiana.1981:384-393.
[36] Mikio S, Yoshifumi Y, Yoshiki Y Y and Masayuki T. Scaling of NDVI and VSWindex between LANDSAT TM and NOAA AVHRR data. Asian Conference OnRemote Sensing.1996.
[37] Cihlar J, Latifovic R, Chen J, Beaubien J and Li Z. Selecting Representative HighResolution Sample Images for Land Cover Studies. Part1Methodology. RemoteSensing of Environment.2000,71:26-42.
[38] Cihlar J. Land cover mapping of large areas from satellites: status and researchpriorities. International Journal of Remote Sensing.2000,21(6&7):1093–1114.
[39] Kellenberger T. Upscaling of spectrally derived land-surface parameters. Int. Symp.On Physical Measurements&Signatures in Remote Sensing.2001:111-116.
[40] Han P, Gong J Y, Li Z L, Bo Y C and Cheng L.2010. Selection of optimal scale inremotely sensed image classification. Journal of Remote Sensing.14(3):507-518.
[41]刘高焕.洪水灾情监测预报中遥感信息复合试验.遥感信息.1984,7.
[42] Liu J Y, Zhuang D F, Luo D et al. Land-cover Classification of China: IntegratedAnalysis of AVHRR Imagery and Geophysical Data. Int. J. of Remote Sensing,2003,24(12):2485-2500.
[43] Estreguil C and Souza G D. Remote sensing measurement of forest parameters in theinsular part of South-East Asia: high/low resolution image data analysis. IGARSS'93.1993.
[44] Cihlar J, Beaubien J, Xiao Q, Chen J M and Li Z. Land cover of the BOREASRegion from AVHRR and Landsat data. Canadian Journal of Remote Sensing.1997,23:163-175.
[45] Laporte N T, Geotz S J, Justice C O and Heinicke M. A new land cover map ofcentral Africa derived from multi-resolution, multi-temporal AVHRR data. INT. J.Remote Sensing.1998,19(18):3537-3550.
[46] Braswell B H, Hagen S C, Frolking S E et al. A multivariable approach for mappingsub-pixel land cover distributions using MISR and MODIS: Application in theBrazilian Amazon region. Remote Sensing of Environment,2003,87(2-3):243-256.
[47]刘闯，葛成辉.美国对地观测系统(EOS)中分辨率成像光谱仪(MODIS)遥感数据的特点与应用.遥感信息.2000,3:45-48.
[48]邢素丽,张广录．我国农业遥感的应用现状与展望[J]．农业工程学报,2003,19(6):174-178.
[49]钱永兰,杨邦杰,雷廷武,数据融合及其在农情遥感监测中的应用与展望[J]．农业工程学报,2004,20(4):286-290.
[50] Matsuoka M, Hayasaka T, Fukushima Y and Honda Y. Analysis of the land coverand its change over Yellow River basin using satellite data. Geo-Imagery BridgingContinents ISPRS.2004.
[51] Braswell B H, Hagen S C, Frolking S E and Salas W A. A multivariable approach formapping sub-pixel land cover distributions using MISR and MODIS: Application inthe Brazilian Amazon region.2003,87(2-3):243-256.
[52] Liu J Y, Zhuang D F, Luo D and Xiao X. Land-cover Classification of China:Integrated Analysis of AVHRR Imagery and Geophysical Data. International Journalof Remote Sensing.2003,24(12):2485-2500.
[53] DeFries R, Hansen M, Townshend J and Sohlberg R. Global land cover classificationat8km spatial resolution: the use of training data derived from Landsat imagery indecision tree classifiers. International Journal of Remote Sensing.1998,19:3141–3168.
[54] Zhu Z and Evans D. U.S. forest cover types and predicted per cent forest cover fromAVHRR data. Photogrammetric Engineering and Remote Sensing.1994,60:525-531.
[55] Fazakas Z and Nilsson M. Volume and forest cover estimation over southern Swedenusing AVHRR data calibrated with TM data. International Journal of RemoteSensing.1996,17(9):1701-1709.
[56] H me T, Salli A, Andersson K and Lohi A. A new methodology for the estimation ofbiomass of conifer dominated boreal forest using NOAA AVHRR data. InternationalJournal of Remote Sensing.1997,18(15):3211-3243.
[57] Tomppo E, Nilsson M, Rosengren M, Aalto P and Kennedy P. Simultaneous use ofLandsat-TM and IRS-1C WiFS data in estimating large area tree stem volume andaboveground biomass. Remote Sensing of Environment.2002,82(1):156-171.
[58]邬明权,王长耀,牛铮.利用多源时序遥感数据提取大范围水稻种植面积[J].农业工程学报,2010,26(7):240-244.
[59] Kasetkasem T, Arora M K and Varshney P K. Sub-pixel Land Cover Mapping Basedon Markov Random Field Models. IEEE Geoscience and Remote SensingSymposium.2003,6:3456-3458.
[60] Verhoeye J and Wulf R D. Land cover mapping at sub-pixel scales using linearoptimization techniques. Remote Sensing of Environment.2002,79(1):96-104.
[61] H me T, Stenberg P and Rauste Y. A methodology to estimate forest variables atsub-pixel level. Conference on Remote Sensing and Forest Monitoring. EuropeanCommunities. Luxembourg.2000:451-461.
[62]秦华.多源遥感图像融合和地表复合分类算法研究：[硕士学位论文].北京：清华大学电子系，2003.
[63]何文举.基于多分辨率遥感图像的复合分类方法研究[D].北京:清华大学硕士学位论文,2006:10-35.
[64]赵春霞，钱乐祥，遥感影像监督分类与非监督分类的比较，河南大学学报(自然科学版)，2004:34(3).
[65] Fazakas Z, Nilsson M. Volume and Forest Cover Estimation over Southern Swedenusing AVHRR Data Calibrated with TM Data. Int. J. of Remote Sensing,1996,17(9):1701-1709.
[66] Tomppo E, Nilsson M, Rosengren M et al. Simultaneous use of Landsat-TM andIRS-1C WiFS data in estimating large area tree stem volume and abovegroundbiomass. Remote Sensing of Environment,2002,82(1):156-171.
[67] Qin H, Li M and Sun W D. A Combined Interpolation for Land Cover Using AVHRRData Supervised by TM Data. IEEE Inter. Conf. on RISSP,2003,920-925.
[68] He W J, Qin H and Sun W D. Compounded Method for Land Covering Classificationbased on Multi-resolution satellite data, proceeding of the International Symposiumon Remote Sensing, Jeju, Korea,2005.
[69] Foody G M. Approaches for the production and evaluation of fuzzy land coverclassifications from remotely-sensed data. INT. J. Remote Sensing.1996,17(7):1317-1340.
[70] Doraiswamy P, Hollinger S, Evaluation of MODIS Products for applications inRegional crop condition and yield assessment. Remote Sensing For AgricultureEcosystems And Hydrology. November17,2001.
[71]焦险峰.大尺度作物面积遥感抽样调查方法研究[D].北京:中国农业大学,2005.
[72] Mumby P J and Edwards A J. Mapping Marine Environments with IKONOSImagery: Enhanced Spatial Resolution Can Deliver Greater Thematic Accuracy.Remote Sensing of Environment.2002,82:248-257.
[73] Ferro, C. J. S. and T. A. Warner. Scale and Texture Digital Image Classification [J].Photogrammetric Engineering&Remote Sensing.2002,68(1):51-63.
[74]刘龙飞,陈云浩,李京.遥感影像纹理分析方法综述与展望.遥感技术与应用.2003,18(6):441-446.
[75]田艳琴,郭平,卢汉清.基于灰度矩阵的多波段遥感图像纹理特征的提取.计算机科学.2004,4(31):162-164.
[76]袁路.基于像元分解与不确定性分析的遥感分类评估方法[硕士学位论文].北京:清华大学电子工程系,2005.
[77]郭山,袁路,孙卫东.基于像元分解的亚像元级遥感数据分类精度评估方法.高技术通讯,2007,17(10):1044-1049.
[78] Foody G M. Status of land cover classification accuracy assessment. RemoteSensing of Environment.2002,80:185-201.
[79] Foody G M. Approaches for the production and evaluation of fuzzy land coverclassifications from remotely-sensed data. INT. J. Remote Sensing.1996,17(7):1317-1340.
[80] Stehman S V. Practical Implications of Design-based Sampling for Thematic MapAccuracy Assessment. Remote Sensing of Environment.2000,72:35-45.
[81] Wu W and Shao G. Optimal Combinations of Data, Classifiers, and Sampling Methodsfor Accurate Characterizations of Deforestation. Canadian Journal of Remote Sensing.2002,28(4):601-609.
[82] Vapnik V N.统计学习理论的本质[M].北京:清华大学出版社.2000.
[83] Congaltom R G and Green K. Asseing the Accuracy of Remotely Sensed Data:Principles and Practices, Boca Raton, FL: Lewis Publishers.1999,137.
[84]陈小玲,龚威,李平湘等(译).遥感数字影像处理导论.北京:机械工业出版社.2007:483.
[85]张良培,张立福.高光谱遥感.武汉:武汉大学出版社,2005.
[86] Members of the MODIS Characterization Support Team For NASA/Goddard SpaceFlight Center. MODIS Level1B Product User's Guide For Level1B Version4.3.0(Terra) and Version4.3.1(Aqua).2003.
[87] Zhan X, Defries R, Townshend J R G, et al. The250m GlobalL and Cover ChangeProduct from the Moderate Resolution Imaging Spectroradiometer of NASA’s EarthObserving System. Journal of Remote Sensing.2000,21:6-7.
[88]刘玉洁,杨忠东. MODIS遥感信息处理原理与算法.北京:科学出版社,2001.
[89] Sykes A O. An Introduction to Regression Analysis. Chicago Working Paper in Law&Economics.1993.
[90]孙九林.中国农作物遥感动态监测与估产总论.北京:中国科学技术出版社.1996.
[91]康晓风,王乃斌,杨小唤.粮食种植面积提取方法的发展与现状.资源科学,2002,24(5):8-12.
[92]郭琳,孙卫东,王琼华,杨邦杰.基于组合核非线性退化模型的多分辨率遥感图像复合分类[J].北京:农业工程学报,2008,24(10):145-150.
[93] Ferro C and Warner T A. Scale and Texture Digital Image Classification.Photogrammetric Engineering&Remote Sensing.2002,68(1):51-63.
[94] Vapink V N. The Nature of Statistical Learning Theory [M]. New York: SpringerVerlag.1995.
[95] Smits G, Jordan E. Improved SVM Regression Using Mixtures of Kernels[C].Honolulu: IEEE Press.2002:2785-2790.
[96] Baraldi A, Parminggian F. An Investigation on the Texture Characteristics Associatedwith Gray Level Co-occurrence Matrix Statistical Parameters [J]. IEEE Trans. OnGeosciences and Remote Sensing.1995,32(2):293-303.
[97]冯嘉,支持向量机分类算法及其在进化计算中的应用研究[D],西安理工大学硕士学位论文,2010,3:3-25.
[98] Perze-Cruz F, Bousquet O. Kernel methods and their potential use in signal processing.IEEE Signal Processing Magazine.2004,21(3):57-65.
[99]陈渤.一种融合核优化算法.西安电子科技大学学报.2007,34(4):509-513.
[100]张冰,孔锐.一种支持向量机的组合核函数.计算机应用.2007,1:44-46.
[101] Xiong H L, Swamy M N S, Ahmad M O. Optimizing the Kernel in the Empiricalfeature Space[J].IEEE Trans on Neural Networks.2005,16(2):460-474.
[102]赵明.多元线性回归预测及其检验在EXCEL中的实现.吉林化工学院学报.2003,2(20):85-87.
[103] Iverson L R, Cook E A, and Graham R L. Regional forest cover estimation viaremote sensing: the calibration center concept. Landscape Ecology9.1994.159-174.
[104] Wang F J. Fuzzy Supervised Classification of Remote Sensing Images. IEEETransaction on Geoscience and Remote Sensing.1990,28(2):194-201.
[105]扎西顿珠,普布次仁.对西藏“一江两河”遥感影像镶嵌并进行拼接缝消除及分类[J].遥感技术与应用,2006,21(5):463-467, I0002.
[106] Anders H. On the History of Maximum Likelihood in Relation to Inverse Probabilityand Least Squares. Statistical Science,1999:214–222.
[107] Qin H, Li M and Sun W D. A Combined Interpolation for Land Cover Using AVHRRData Supervised by TM Data. IEEE Inter. Conf. on RISSP.2003:920-925.
[108] Vapnik V N and Mukherjee S. Support Vector Method for Multivariate DensityEstimation. Advances in Neural Information Processing Systems.1999:659-665.
[109] Chen D. and Stow D. Strategies for integrating information from multiple spatialresolutions into land use/cover classification routines. Photogrammetric Engineering&Remote Sensing,2003,69(11): pp.1279-1287.
[110]李兴生,基于云模型和数据场的分类和聚类挖掘研究,中国人民解放军理工大学博士学位论文,2003.
[111] Zhong P and Wang R S. A Multiple Conditional Random Fields Ensemble Model forUrban Area Detection in Remote Sensing Optical Images. IEEE Trans. OnGeoscience and Remote Sensing,2007,45(12):3978-3988.
[112]褚一平.基于条件随机场模型的视频目标分割算法研究,浙江大学博士学位论文.2007.12:5-20.
[113] Lafferty J, McCallum A, and Pereira F. Conditional Random Fields: ProbabilisticModels for Segmenting and Labeling Sequence Data. In ICML,2001.
[114] Max W and Sridevi P. Bayesian Random Fields: the Bethe-Laplace Approximation.Uncertainty in Artificial Intelligence,2006.
[115] Kengo S, and Yasubumi S. RNA Secondary Structural alignment with conditionalrandom fields. Bioinformatics,2005,21:237-242.
[116] Burr S. Abner:an Open Source Tool for Automatically Tagging Genes, Proteins, andOther Entity Names in Text. Bioinformatics,2005,21(14):3191-3192.
[117] Wainwright M J and Jordan M I. Graphical Models, Exponential Families, andVariational Inference. Foundations and Trends in Machine Learning.2008,1(1-2):1-305.
[118] Hanna W. Efficient Training of Conditional Random Fields. M.Sc. thesis, Universityof Edinburgh,2002.
[119] CRF++, http://crfpp.sourceforge.net/.
[120]王乃斌.中国小麦遥感动态监测与估产.北京:中国科学技术出版社.1996.
[121]王茂新,裴志远,吴全,杨邦杰.应用NOAA图像监测冬小麦面积的研究.农业工程学报,1999,15(2):181-184.
[122]杨邦杰,裴志远.国家级农情遥感监测系统的研究、开发与运行.农业工程学报.2003,19(增刊):11-14.
[123]刘海启,金敏毓,龚维鹏.美国农业遥感技术应用状况概述.中国农业资源与区别,1999,20(2).
[124]周清波.国内外农情遥感现状与发展趋势.中国农业资源与区划,2004,25(5):9-14.
[125]全国冬小麦遥感综合测产协作组.冬小麦气象卫星遥感动态监测与估产.北京：气象出版社.1993.
[126]杨邦杰,陆登槐,裴志远等.国家级农情监测系统结构设计.农业工程学报,1997,13(1):16-19.
[127]郭琳,基于多分辨率遥感数据的作物复合分类方法[硕士学位论文].北京:清华大学电子工程系,2008.
[2] Gat N, Subramanian S, Barhen J, et al. Spectral Imaging Applications: RemoteSensing, Environmental Monitoring, Medicine, Military Operations, FactoryAutomation and Manufacturing. Proceeding of SPIE on Emerging Applications ofComputer Vision, Volume2962. The International Society for Optical Engineering,1997:63-77.
[3]朱述龙,朱宝山,王红卫.遥感图像处理与应用.北京:科学出版社,2006.
[4] Bauer, M E. Technological Basis and Applications of Remote Sensing of the Earth'sResources. IEEE Transactions on Geoscience Electronics,1976,14(1):3-9.
[5] Du P J, Zhang H P, Pan C, et al. Applications of Multi-source Remote SensingInformation to Urban Environment Monitoring in Mining Industrial Cities. UrbanRemote Sensing Joint Event,2007:1-12.
[6]钱永兰.运行化作物遥感识别中的多源数据组织与融合方法研究[D].北京:中国科学院,2005:1-13.
[7]孙卫东.基于多分辨率遥感图像的复合分类方法研究[国家自然科学基金结题报告(课题编号:60472029)].北京:清华大学电子系,2008.
[8]熊育久,林辉,孙华,等.多源遥感数据融合及其对植被识别的影响.林业资源管理,2005(5):71-76.
[9]李军,周月琴,李德仁.小波变换用于高分辨率全色影像与多光谱影像的融合研究.遥感学报,1999,3(2):116-120.
[10]陈思锦,秦其明,王文君.利用小波分析改进Brovey遥感影像融合方法.测绘学院学报,2004,21(2):118-120.
[11]董毓敏.基于小波叠加的主成分变换遥感数据融合方法的研究.东北测绘,2002,25(3):10-11.
[12]张炳智,张继贤,张丽.土地利用动态遥感监测中多源遥感影像融合方法比较研究.测绘科学,2000(3):46-51.
[13]贾永红,孙家柄.遥感多光谱影像数据与航片数字化影像融合方法的研究.测绘通报,1997(5):10-12.
[14]金剑秋,王章野,江照意,等.多光谱图像的真实感融合.中国图像图形学报,2002,7(A)(9):926-931.
[15]王智均,李德仁,李清泉.利用小波变换对影像进行融合的研究.武汉测绘科技大学学报,2000,25(2):137-142.
[16]张炳智,张继贤,张丽.土地利用动态遥感监测中多源遥感影像融合方法比较研究.测绘科学,2000(3):46-51.
[17] Tax T, Solberg S. Information Fusion in Remote Sensing. Vista in Astronomy,1997,41(3):337-342.
[18] Ronald Y. A framework for multi-source data fusion. Information Sciences,2004,163:175-200.
[19] Landgrebe D. On the relationship between class definition precision andclassification accuracy in hyperspectral analysis. IGARSS,2000:147-149.
[20] Harsanyi J, Chang C. Hyperspectral image classification and dimensionalityreduction：an orthogonal subspace projection approach. IEEE Trans. on G. R. S.1994,32(4):779-785.
[21]蔡学良,崔远来.基于异源多时相遥感数据提取灌区作物种植结构.北京:农业工程学报,2009,25(8):124-131.
[22] Jia X, Richards J. Segmented principal components transformation for efficienthyperspectral remote sensing image display and classification. IEEE Trans. On G. R.S.1999,37(1):538-542.
[23]张钧萍,张晔.基于多特征多分辨率融合的高光谱图像分类.红外与毫米波学报.2004,23(5):345-348.
[24] Zhang Y, Desai M, Zhang J, et al. Adaptive subspace decomposition forhyperspectral data dimensionality reduction. ICIP.1999:326-329.
[25]黄宁，朱敏慧，张守融，一种采用高斯隐马尔可夫随机场模型的遥感图像分类算法，电子与信息学报,2003,25(1):50-53.
[26]郭琳.基于多分辨率遥感数据的作物复合分类方法.北京:清华大学硕士学位论文,2008.
[27] DeFries R, Hansen M, Steininger M, Dubayah R, Sohlberg R and Townshend J.Subpixel forest cover in central Africa from multisensor, multitemporal data.Remote Sensing of Environment.1997,60:228–246.
[28]强强.不同分辨率遥感图像在地表分类中的综合应用.北京:清华大学电子系毕业设计论文,2003.
[29] Zhukov B, Oertel D, and Lanzl F. A multiresolution multisensor technique forsatellite remote sensing. IGARSS,1995,51-53.
[30] Cihlar J, Latifovic R, Chen J, Beaubien J and Li Z. Selecting Representative HighResolution Sample Images for Land Cover Studies. Part2: Application to EstimatingLand Cover Composition. Remote Sensing of Environment.2000,72:127-138.
[31] Rokhmatuloh, Bilbisi H A, Arihara K, et al. Application of Regression Tree Methodfor Estimating Percent Tree Cover of Asia with QuickBird images as training data.Proceedings of the CEReS International Symposium on Remote Sensing.2005.
[32]钱永兰.运行化作物遥感识别中的多源数据组织与融合方法研究[D].北京:中国科学院硕士学位论文,2005:1-13.
[33] Cushnie J L. The interactive effect of spatial resolution and degree of internalvariability within land-cover types on classification accuracies. International Journalof Remote Sensing,1987,8(1):15-29.
[34] Chen D M A multi-resolution analysis and classification framework for improvingland use/cover mapping from earth observation data. The International Archives ofthe Photogrammetry, Remote Sensing and Spatial Information Sciences,2004,34.
[35] Latty R S and Hoffer R M. Computer based classification accuracy due to the spatialresolution using per-point versus per-field classification techniques. Proc. ofRemotely Sensed Data Symposium, West Lafayette, Indiana.1981:384-393.
[36] Mikio S, Yoshifumi Y, Yoshiki Y Y and Masayuki T. Scaling of NDVI and VSWindex between LANDSAT TM and NOAA AVHRR data. Asian Conference OnRemote Sensing.1996.
[37] Cihlar J, Latifovic R, Chen J, Beaubien J and Li Z. Selecting Representative HighResolution Sample Images for Land Cover Studies. Part1Methodology. RemoteSensing of Environment.2000,71:26-42.
[38] Cihlar J. Land cover mapping of large areas from satellites: status and researchpriorities. International Journal of Remote Sensing.2000,21(6&7):1093–1114.
[39] Kellenberger T. Upscaling of spectrally derived land-surface parameters. Int. Symp.On Physical Measurements&Signatures in Remote Sensing.2001:111-116.
[40] Han P, Gong J Y, Li Z L, Bo Y C and Cheng L.2010. Selection of optimal scale inremotely sensed image classification. Journal of Remote Sensing.14(3):507-518.
[41]刘高焕.洪水灾情监测预报中遥感信息复合试验.遥感信息.1984,7.
[42] Liu J Y, Zhuang D F, Luo D et al. Land-cover Classification of China: IntegratedAnalysis of AVHRR Imagery and Geophysical Data. Int. J. of Remote Sensing,2003,24(12):2485-2500.
[43] Estreguil C and Souza G D. Remote sensing measurement of forest parameters in theinsular part of South-East Asia: high/low resolution image data analysis. IGARSS'93.1993.
[44] Cihlar J, Beaubien J, Xiao Q, Chen J M and Li Z. Land cover of the BOREASRegion from AVHRR and Landsat data. Canadian Journal of Remote Sensing.1997,23:163-175.
[45] Laporte N T, Geotz S J, Justice C O and Heinicke M. A new land cover map ofcentral Africa derived from multi-resolution, multi-temporal AVHRR data. INT. J.Remote Sensing.1998,19(18):3537-3550.
[46] Braswell B H, Hagen S C, Frolking S E et al. A multivariable approach for mappingsub-pixel land cover distributions using MISR and MODIS: Application in theBrazilian Amazon region. Remote Sensing of Environment,2003,87(2-3):243-256.
[47]刘闯，葛成辉.美国对地观测系统(EOS)中分辨率成像光谱仪(MODIS)遥感数据的特点与应用.遥感信息.2000,3:45-48.
[48]邢素丽,张广录．我国农业遥感的应用现状与展望[J]．农业工程学报,2003,19(6):174-178.
[49]钱永兰,杨邦杰,雷廷武,数据融合及其在农情遥感监测中的应用与展望[J]．农业工程学报,2004,20(4):286-290.
[50] Matsuoka M, Hayasaka T, Fukushima Y and Honda Y. Analysis of the land coverand its change over Yellow River basin using satellite data. Geo-Imagery BridgingContinents ISPRS.2004.
[51] Braswell B H, Hagen S C, Frolking S E and Salas W A. A multivariable approach formapping sub-pixel land cover distributions using MISR and MODIS: Application inthe Brazilian Amazon region.2003,87(2-3):243-256.
[52] Liu J Y, Zhuang D F, Luo D and Xiao X. Land-cover Classification of China:Integrated Analysis of AVHRR Imagery and Geophysical Data. International Journalof Remote Sensing.2003,24(12):2485-2500.
[53] DeFries R, Hansen M, Townshend J and Sohlberg R. Global land cover classificationat8km spatial resolution: the use of training data derived from Landsat imagery indecision tree classifiers. International Journal of Remote Sensing.1998,19:3141–3168.
[54] Zhu Z and Evans D. U.S. forest cover types and predicted per cent forest cover fromAVHRR data. Photogrammetric Engineering and Remote Sensing.1994,60:525-531.
[55] Fazakas Z and Nilsson M. Volume and forest cover estimation over southern Swedenusing AVHRR data calibrated with TM data. International Journal of RemoteSensing.1996,17(9):1701-1709.
[56] H me T, Salli A, Andersson K and Lohi A. A new methodology for the estimation ofbiomass of conifer dominated boreal forest using NOAA AVHRR data. InternationalJournal of Remote Sensing.1997,18(15):3211-3243.
[57] Tomppo E, Nilsson M, Rosengren M, Aalto P and Kennedy P. Simultaneous use ofLandsat-TM and IRS-1C WiFS data in estimating large area tree stem volume andaboveground biomass. Remote Sensing of Environment.2002,82(1):156-171.
[58]邬明权,王长耀,牛铮.利用多源时序遥感数据提取大范围水稻种植面积[J].农业工程学报,2010,26(7):240-244.
[59] Kasetkasem T, Arora M K and Varshney P K. Sub-pixel Land Cover Mapping Basedon Markov Random Field Models. IEEE Geoscience and Remote SensingSymposium.2003,6:3456-3458.
[60] Verhoeye J and Wulf R D. Land cover mapping at sub-pixel scales using linearoptimization techniques. Remote Sensing of Environment.2002,79(1):96-104.
[61] H me T, Stenberg P and Rauste Y. A methodology to estimate forest variables atsub-pixel level. Conference on Remote Sensing and Forest Monitoring. EuropeanCommunities. Luxembourg.2000:451-461.
[62]秦华.多源遥感图像融合和地表复合分类算法研究：[硕士学位论文].北京：清华大学电子系，2003.
[63]何文举.基于多分辨率遥感图像的复合分类方法研究[D].北京:清华大学硕士学位论文,2006:10-35.
[64]赵春霞，钱乐祥，遥感影像监督分类与非监督分类的比较，河南大学学报(自然科学版)，2004:34(3).
[65] Fazakas Z, Nilsson M. Volume and Forest Cover Estimation over Southern Swedenusing AVHRR Data Calibrated with TM Data. Int. J. of Remote Sensing,1996,17(9):1701-1709.
[66] Tomppo E, Nilsson M, Rosengren M et al. Simultaneous use of Landsat-TM andIRS-1C WiFS data in estimating large area tree stem volume and abovegroundbiomass. Remote Sensing of Environment,2002,82(1):156-171.
[67] Qin H, Li M and Sun W D. A Combined Interpolation for Land Cover Using AVHRRData Supervised by TM Data. IEEE Inter. Conf. on RISSP,2003,920-925.
[68] He W J, Qin H and Sun W D. Compounded Method for Land Covering Classificationbased on Multi-resolution satellite data, proceeding of the International Symposiumon Remote Sensing, Jeju, Korea,2005.
[69] Foody G M. Approaches for the production and evaluation of fuzzy land coverclassifications from remotely-sensed data. INT. J. Remote Sensing.1996,17(7):1317-1340.
[70] Doraiswamy P, Hollinger S, Evaluation of MODIS Products for applications inRegional crop condition and yield assessment. Remote Sensing For AgricultureEcosystems And Hydrology. November17,2001.
[71]焦险峰.大尺度作物面积遥感抽样调查方法研究[D].北京:中国农业大学,2005.
[72] Mumby P J and Edwards A J. Mapping Marine Environments with IKONOSImagery: Enhanced Spatial Resolution Can Deliver Greater Thematic Accuracy.Remote Sensing of Environment.2002,82:248-257.
[73] Ferro, C. J. S. and T. A. Warner. Scale and Texture Digital Image Classification [J].Photogrammetric Engineering&Remote Sensing.2002,68(1):51-63.
[74]刘龙飞,陈云浩,李京.遥感影像纹理分析方法综述与展望.遥感技术与应用.2003,18(6):441-446.
[75]田艳琴,郭平,卢汉清.基于灰度矩阵的多波段遥感图像纹理特征的提取.计算机科学.2004,4(31):162-164.
[76]袁路.基于像元分解与不确定性分析的遥感分类评估方法[硕士学位论文].北京:清华大学电子工程系,2005.
[77]郭山,袁路,孙卫东.基于像元分解的亚像元级遥感数据分类精度评估方法.高技术通讯,2007,17(10):1044-1049.
[78] Foody G M. Status of land cover classification accuracy assessment. RemoteSensing of Environment.2002,80:185-201.
[79] Foody G M. Approaches for the production and evaluation of fuzzy land coverclassifications from remotely-sensed data. INT. J. Remote Sensing.1996,17(7):1317-1340.
[80] Stehman S V. Practical Implications of Design-based Sampling for Thematic MapAccuracy Assessment. Remote Sensing of Environment.2000,72:35-45.
[81] Wu W and Shao G. Optimal Combinations of Data, Classifiers, and Sampling Methodsfor Accurate Characterizations of Deforestation. Canadian Journal of Remote Sensing.2002,28(4):601-609.
[82] Vapnik V N.统计学习理论的本质[M].北京:清华大学出版社.2000.
[83] Congaltom R G and Green K. Asseing the Accuracy of Remotely Sensed Data:Principles and Practices, Boca Raton, FL: Lewis Publishers.1999,137.
[84]陈小玲,龚威,李平湘等(译).遥感数字影像处理导论.北京:机械工业出版社.2007:483.
[85]张良培,张立福.高光谱遥感.武汉:武汉大学出版社,2005.
[86] Members of the MODIS Characterization Support Team For NASA/Goddard SpaceFlight Center. MODIS Level1B Product User's Guide For Level1B Version4.3.0(Terra) and Version4.3.1(Aqua).2003.
[87] Zhan X, Defries R, Townshend J R G, et al. The250m GlobalL and Cover ChangeProduct from the Moderate Resolution Imaging Spectroradiometer of NASA’s EarthObserving System. Journal of Remote Sensing.2000,21:6-7.
[88]刘玉洁,杨忠东. MODIS遥感信息处理原理与算法.北京:科学出版社,2001.
[89] Sykes A O. An Introduction to Regression Analysis. Chicago Working Paper in Law&Economics.1993.
[90]孙九林.中国农作物遥感动态监测与估产总论.北京:中国科学技术出版社.1996.
[91]康晓风,王乃斌,杨小唤.粮食种植面积提取方法的发展与现状.资源科学,2002,24(5):8-12.
[92]郭琳,孙卫东,王琼华,杨邦杰.基于组合核非线性退化模型的多分辨率遥感图像复合分类[J].北京:农业工程学报,2008,24(10):145-150.
[93] Ferro C and Warner T A. Scale and Texture Digital Image Classification.Photogrammetric Engineering&Remote Sensing.2002,68(1):51-63.
[94] Vapink V N. The Nature of Statistical Learning Theory [M]. New York: SpringerVerlag.1995.
[95] Smits G, Jordan E. Improved SVM Regression Using Mixtures of Kernels[C].Honolulu: IEEE Press.2002:2785-2790.
[96] Baraldi A, Parminggian F. An Investigation on the Texture Characteristics Associatedwith Gray Level Co-occurrence Matrix Statistical Parameters [J]. IEEE Trans. OnGeosciences and Remote Sensing.1995,32(2):293-303.
[97]冯嘉,支持向量机分类算法及其在进化计算中的应用研究[D],西安理工大学硕士学位论文,2010,3:3-25.
[98] Perze-Cruz F, Bousquet O. Kernel methods and their potential use in signal processing.IEEE Signal Processing Magazine.2004,21(3):57-65.
[99]陈渤.一种融合核优化算法.西安电子科技大学学报.2007,34(4):509-513.
[100]张冰,孔锐.一种支持向量机的组合核函数.计算机应用.2007,1:44-46.
[101] Xiong H L, Swamy M N S, Ahmad M O. Optimizing the Kernel in the Empiricalfeature Space[J].IEEE Trans on Neural Networks.2005,16(2):460-474.
[102]赵明.多元线性回归预测及其检验在EXCEL中的实现.吉林化工学院学报.2003,2(20):85-87.
[103] Iverson L R, Cook E A, and Graham R L. Regional forest cover estimation viaremote sensing: the calibration center concept. Landscape Ecology9.1994.159-174.
[104] Wang F J. Fuzzy Supervised Classification of Remote Sensing Images. IEEETransaction on Geoscience and Remote Sensing.1990,28(2):194-201.
[105]扎西顿珠,普布次仁.对西藏“一江两河”遥感影像镶嵌并进行拼接缝消除及分类[J].遥感技术与应用,2006,21(5):463-467, I0002.
[106] Anders H. On the History of Maximum Likelihood in Relation to Inverse Probabilityand Least Squares. Statistical Science,1999:214–222.
[107] Qin H, Li M and Sun W D. A Combined Interpolation for Land Cover Using AVHRRData Supervised by TM Data. IEEE Inter. Conf. on RISSP.2003:920-925.
[108] Vapnik V N and Mukherjee S. Support Vector Method for Multivariate DensityEstimation. Advances in Neural Information Processing Systems.1999:659-665.
[109] Chen D. and Stow D. Strategies for integrating information from multiple spatialresolutions into land use/cover classification routines. Photogrammetric Engineering&Remote Sensing,2003,69(11): pp.1279-1287.
[110]李兴生,基于云模型和数据场的分类和聚类挖掘研究,中国人民解放军理工大学博士学位论文,2003.
[111] Zhong P and Wang R S. A Multiple Conditional Random Fields Ensemble Model forUrban Area Detection in Remote Sensing Optical Images. IEEE Trans. OnGeoscience and Remote Sensing,2007,45(12):3978-3988.
[112]褚一平.基于条件随机场模型的视频目标分割算法研究,浙江大学博士学位论文.2007.12:5-20.
[113] Lafferty J, McCallum A, and Pereira F. Conditional Random Fields: ProbabilisticModels for Segmenting and Labeling Sequence Data. In ICML,2001.
[114] Max W and Sridevi P. Bayesian Random Fields: the Bethe-Laplace Approximation.Uncertainty in Artificial Intelligence,2006.
[115] Kengo S, and Yasubumi S. RNA Secondary Structural alignment with conditionalrandom fields. Bioinformatics,2005,21:237-242.
[116] Burr S. Abner:an Open Source Tool for Automatically Tagging Genes, Proteins, andOther Entity Names in Text. Bioinformatics,2005,21(14):3191-3192.
[117] Wainwright M J and Jordan M I. Graphical Models, Exponential Families, andVariational Inference. Foundations and Trends in Machine Learning.2008,1(1-2):1-305.
[118] Hanna W. Efficient Training of Conditional Random Fields. M.Sc. thesis, Universityof Edinburgh,2002.
[119] CRF++, http://crfpp.sourceforge.net/.
[120]王乃斌.中国小麦遥感动态监测与估产.北京:中国科学技术出版社.1996.
[121]王茂新,裴志远,吴全,杨邦杰.应用NOAA图像监测冬小麦面积的研究.农业工程学报,1999,15(2):181-184.
[122]杨邦杰,裴志远.国家级农情遥感监测系统的研究、开发与运行.农业工程学报.2003,19(增刊):11-14.
[123]刘海启,金敏毓,龚维鹏.美国农业遥感技术应用状况概述.中国农业资源与区别,1999,20(2).
[124]周清波.国内外农情遥感现状与发展趋势.中国农业资源与区划,2004,25(5):9-14.
[125]全国冬小麦遥感综合测产协作组.冬小麦气象卫星遥感动态监测与估产.北京：气象出版社.1993.
[126]杨邦杰,陆登槐,裴志远等.国家级农情监测系统结构设计.农业工程学报,1997,13(1):16-19.
[127]郭琳,基于多分辨率遥感数据的作物复合分类方法[硕士学位论文].北京:清华大学电子工程系,2008.