·404· 智能系统学报 第6卷 ence on Machine Learning.Washington,DC,USA, [41]ZHONG Ping,WANG Runsheng.A multiple conditional 2003:912914. random fields ensemble model for urban area detection in [30]SINDHWANI V,CHU W,KEERTHI S S.Semi-super- remote sensing optical images[J].IEEE Transactions on vised Gaussian process classifiers[C]//Interational Joint Geoscience and Remote Sensing,2007,45(12):3978- Conference on Artificial Intelligence.Hyderabad,India, 3988. 2007:1059-1064. [42]LAFFERTY J,MCCALLUM A,PEREIRA F.Conditional [31]HUGHES G.On the mean accuracy of statistical pattem random fields:probabilistic models for segmenting and la- recognizers[J].IEEE Transactions on Information Theo- beling sequence data[C]//Proceedings of the Eighteenth y,1968,14(1):5563. Intemational Conference on Machine Leaming.Williams- [32]QIAN Y,YAO F,JIA S.Band selection for hyperspectral towm,USA,2001:282-289. imagery using affinity propagation[J].IET Computer Vi- [43 KUMAR S,HEBERT M.Discriminative random fields sion,2010,3(4):213-222. [J].International Joumal of Computer Vision,2006,68 [33]贺霖，潘泉，邸韦华，等.高光谱图像高维多尺度自回归 (2):179-201. 有监督检测[J].自动化学报，2009,35(5)：509-518. [44]LI Jiming,HU Zhenfang,QIAN Yuntao.Hyperspectral HE Lin,PAN Quan,DI Weihua,et al.Supervised detec- data classification using margin infused relaxed algorithm tion for hyperspectral imagery based on high dimensional [C]//Interational Conference on Image Processing. multiscale autoregression[J].Acta Automatica Sinica, Hong Kong,China,2009:1669-1672. 2009,35(5):509-518. [45 ]LI Jiming,QIAN Yuntao.Regularized multinomial regres- [34]熊桢，童庆禧.用于高光谱遥感图象分类的一种高阶神 sion method for hyperspectral data classification via path- 经网络算法[J].中国图象图形学报，2000,5(3)： wise coordinate optimization[C]//Digital Image Compu- 196-201. ting:Techniques and Applications.Melbourne,Australia, XIONG Zhen,TONG Qingxi.High-rank artificial neural 2009:540-545. network algorithm for classification of hyperspectral image [46]YAO Futian,Qian Yuntao.Band selection based Gaussian data[J].Joumal of Image and Graphics,2000,5(3): processes for hyperspectral remote sensing images classifi- 196-201. cation[C]//Intemational Conference on Image Processing. [35]MELGANI F,BRUZZONE L.Classification of hyperspec- Hong Kong,China,2009:2845-2848. tral remote sensing images with support vector machines [47]VATSAVAI RR,SHEKHAR S,BURK T E.A semi-su- [J].IEEE Transactions on Geoscience and Remote Sens- pervised learing method for remote sensing data mining ing,2004,42(8):1778-1790. [C]//International Conference on Tools with Artificial In- [36]刘春红.超光谱遥感图像降维及分类方法研究[D].哈 telligence.Hong Kong,China,2005:205-211. 尔滨：哈尔滨工程大学，2005：8694. [48]BELKIN M,NIYOGI P,SINDHWANI V.Manifold regu- LIU Chunhong.Research on dimensional reduction and larization:a geometric framework for learning from labeled classification of hyperspectral remote sensing image[D]. and unlabeled examples[J].The Joumal of Machine Harbin:Harbin Engineering University,2005:86-94. Learing Research,2006,7:2399-2434. [37]KITTLER J,PAIRMAN D.Contextual patter recognition 作者简介： applied to cloud detection and identification [J].IEEE 姚伏天，男，1976年生，博士研究 Transactions on Geoscience and Remote Sensing,2007,23 生，主要研究方向为模式识别、机器学 (6):855863 习、高光谱成像信息处理，发表学术论 [38]姚伏天，钱沄涛.用于高光谱遥感图像分类的空间约束 文20余篇. 高斯过程方法[J].南京大学学报：自然科学版，2009， 45(5):665670. YAO Futian,QIAN Yuntao.A spatial Gaussian process method for hyperspectral remote sensing imagery classifica- tion[J].Journal of Nanjing University:Natural Sciences, 钱沄涛，男，1968年生，教授，博士 2009,45(5):665670. 生导师，中国计算机学会人工智能与模 [39]ROSSI R E,DUNGAN J L,BECK L R.Kriging in the 式识别专业委员会委员、模糊逻辑与多 shadows:geostatistical interpolation for remote sensing 值逻辑专业委员会委员.主要研究方向 [J].Remote Sensing of Environment,1994,49(1):32- 为模式识别、机器学习、信号处理，承担 40. 多项国家自然科学基金项目、国际合作 [40]DENG H,CLAUSI D A.Advanced Gaussian MRF rota- 基金项目和省部级重点科技项目，发表 tion-invariant texture features for classification of remote 学术论文70余篇， sensing imagery [C]//Computer Society Conference on Computer Vision and Patter Recognition.Madison,USA, 2003:685-689