面向图像标记的条件随机场模型研究
|
摘要
高精度的图像标记这一技术难题制约了图像检索的应用水平，而表达与利用图像空间上下文及语义信息是有效解决这一难题的关键技术之一。条件随机场（CRFs）是标记和分割结构化数据的概率框架，它在表达大尺度的图像空间上下文和语义特征以及后验概率建模这两方面有其独特的优势，基于这一概率框架可以有效地降低图像标记中的不确定性，提高图像标记的精度。
基于CRFs理论的上述优势，本文针对面向标记的目标类图像分割中仍存在的问题，开展了考虑目标多尺度特性及语义特征的二阶/高阶CRFs模型的研究。研究的主要内容及创新点如下：
针对目标类图像分割中，如何更好地表达目标空间上下文及语义信息这一问题，本文提出了借助于支持向量机（SVM）将概率潜在语义分析（PLSA）技术与CRFs模型框架有机集成的方法，以实现多类标记下高精度的图像分割。具体地说，将PLSA作为一种特征变换技术衍生出的目标语义特征来作为SVM分类器的输入，并利用此SVM的概率输出来对CRFs模型的关联势函数进行定义，用相邻超像素的特征对比度函数及Potts模型对交互势函数进行了定义，从而建立了一个基于PLSA技术的二阶CRFs模型（简称为PLSA-CRFs）。其中，在PLSA模型建立过程中引进了一种高效的期望最大化（EM）算法来实现PLSA监督参数估计，以改善标准EM算法参数估计效率低的问题。最后，开展了PLSA-CRFs模型的分段参数训练，以及基于环状置信传播（LBP）的模型近似推理方法研究。
针对“单一尺度的CRFs模型不能描述图像目标多层次的空间结构及语义关系”的问题，提出了一种融合目标多级空间上下文及语义关系的二阶RF-CRFs模型以实现高精度的目标类图像分割。具体内容包括：开展了目标分层特征提取及特征提升方法的研究；提出了融合目标三个层级空间上下文及目标语义关系的二阶RF-CRFs模型，其中，该模型是以随机森林（RF）分类器来定义其关联势函数，以多级特征对比度加权的Potts函数来定义交互势函数。最后，开展了该RF-CRFs模型的极大伪似然参数训练，以及基于环状置信传播算法的模型推理方法研究。
针对“基于像素的二阶CRFs模型图像标记效率低”、“基于超像素的二阶CRFs模型中分割尺度及图像过分割质量对标记结果影响较大”以及“二阶CRFs模型空间交互能力有限”的问题，开展了基于标记一致性软约束和考虑图像过分割质量的高阶CRFs模型的创新性研究。具体内容包括：开展了上述模型在多目标类分割问题中的实验研究，并在高阶势函数的定义中，提出了一种面向标记的分割质量敏感性函数的定义方法。最后，对上述高阶CRFs模型开展了启发式分段学习的参数训练方法，以及基于变换的α-膨胀推理方法的研究。
在实验中，以自然图像为测试数据，开展了上述三类模型及其相关标记方法的测试及性能评价，证明了上述方法在复杂场景的多目标图像标记中的有效性。
It is a difficulty of image labelling ensuring high precision that prevents the rapidextension to so many applications such as image retrieval, the expression and modelling ofspatial context and semantic information from a given image is a major key technology tosolve the difficulty in image labelling. Conditional Random Fields（CRFs）is a probabiliticsframework for labeling and segmenting structural data. In the application of Image processing,CRFs have the unique advantages over large scale spatial dependency, semantic informationand directly modelling posteri probability of a given image. Thus, the framework is capable toreduce uncertainty and improve the precision in image labelling.
Owe to the advantages mentioned above of CRFs theory, for some problems existing inobject class image segmentation based on image labelling, the thesis develops the studies onlow and higher order CRFs model when considering multi-scale intrinsic properties andsemantic feature of image, the details and innovative studies are in the following:
For the purpose of better express spatial, even semantic information of image, it isproposed that Probability Latent Semantic Analysis（PLSA）technology has been integratedinto the framework of CRFs via Support Vector Machine(SVM) classifier in order to achievean improved precision of object class segmentation in multi-label case. Details are as follows:Firstly, semantic features were derived from a PLSA clustering procedure can be an input toSVM classifier. Then, an output in the form of probability value from the well-defined SVMclassifier was used to make the associative potential function, Potts function and featurecontrast function were used to define the interactive potential function, then a PLSA-CRFscan be formed. In the process, an efficient Expectation Maximization（EM） algorithm wasintroduced to better estimate the parameters of PLSA model in order to improve less efficientproblem using standard EM algorithm. Finally, piecewise training algorithm is used toestimate the parameters of the proposed model, and Loopy Belief Propagation （LBP）algorithm is used to approximately infer the model.
For the problem “superpixel-based CRFs using features only from a scale can’t expressinherent multi-level spatial structure and semantic relation of an object”, the thesis develops apairwise superpixel-based CRFs model with association term defined as an output of RandomForest classifier based on multi-level spatial context and semantic features of objects so as toacquire a high precision in object class image segmentation. Main studies include: multi-levelspatial context feature extraction and boosted feature selection method, the definition ofassociative potential using Random Forest, and interaction potential weighted by common boundary of neighbors for CRFs model. Finally, pseudo-maximum likelihood method wasused to train the model for parameters estimation, and LBP algorithm was used to infer thepiecewise model.
For the problem of “the pixel-based CRFs is low efficient”,“the scale and quality ofimage over-segmentation have greater effects on superpixel-based CRFs, which have limitedability of spatial interaction” for image labeling, the thesis carried out the studies and makesimprovements on higher order CRFs model based on label consistency soft constraint andquality of image over-segmentation proposed by Kohli. Details are as follows: the model waswell studied by experiments on complex scenes at the first. A new segmentation qualitysensitive function for the purpose of image labelling (not image over-segmentation) was thenproposed, thus higher order potential was updated. Finally, the improved HoCRF in the thesisis trained by piecewise learning algorithm and inferred by transformed α-expansion based ongraph cut method.
Experiments on images including natural, aerial imagery with complex scene are carriedto verify the three proposed models and some relative methods. The results have shown thosemethods are effective to object class segmentation in multi-label case for those images such ascomplex natural scene and aerial images.
基于CRFs理论的上述优势，本文针对面向标记的目标类图像分割中仍存在的问题，开展了考虑目标多尺度特性及语义特征的二阶/高阶CRFs模型的研究。研究的主要内容及创新点如下：
针对目标类图像分割中，如何更好地表达目标空间上下文及语义信息这一问题，本文提出了借助于支持向量机（SVM）将概率潜在语义分析（PLSA）技术与CRFs模型框架有机集成的方法，以实现多类标记下高精度的图像分割。具体地说，将PLSA作为一种特征变换技术衍生出的目标语义特征来作为SVM分类器的输入，并利用此SVM的概率输出来对CRFs模型的关联势函数进行定义，用相邻超像素的特征对比度函数及Potts模型对交互势函数进行了定义，从而建立了一个基于PLSA技术的二阶CRFs模型（简称为PLSA-CRFs）。其中，在PLSA模型建立过程中引进了一种高效的期望最大化（EM）算法来实现PLSA监督参数估计，以改善标准EM算法参数估计效率低的问题。最后，开展了PLSA-CRFs模型的分段参数训练，以及基于环状置信传播（LBP）的模型近似推理方法研究。
针对“单一尺度的CRFs模型不能描述图像目标多层次的空间结构及语义关系”的问题，提出了一种融合目标多级空间上下文及语义关系的二阶RF-CRFs模型以实现高精度的目标类图像分割。具体内容包括：开展了目标分层特征提取及特征提升方法的研究；提出了融合目标三个层级空间上下文及目标语义关系的二阶RF-CRFs模型，其中，该模型是以随机森林（RF）分类器来定义其关联势函数，以多级特征对比度加权的Potts函数来定义交互势函数。最后，开展了该RF-CRFs模型的极大伪似然参数训练，以及基于环状置信传播算法的模型推理方法研究。
针对“基于像素的二阶CRFs模型图像标记效率低”、“基于超像素的二阶CRFs模型中分割尺度及图像过分割质量对标记结果影响较大”以及“二阶CRFs模型空间交互能力有限”的问题，开展了基于标记一致性软约束和考虑图像过分割质量的高阶CRFs模型的创新性研究。具体内容包括：开展了上述模型在多目标类分割问题中的实验研究，并在高阶势函数的定义中，提出了一种面向标记的分割质量敏感性函数的定义方法。最后，对上述高阶CRFs模型开展了启发式分段学习的参数训练方法，以及基于变换的α-膨胀推理方法的研究。
在实验中，以自然图像为测试数据，开展了上述三类模型及其相关标记方法的测试及性能评价，证明了上述方法在复杂场景的多目标图像标记中的有效性。
It is a difficulty of image labelling ensuring high precision that prevents the rapidextension to so many applications such as image retrieval, the expression and modelling ofspatial context and semantic information from a given image is a major key technology tosolve the difficulty in image labelling. Conditional Random Fields（CRFs）is a probabiliticsframework for labeling and segmenting structural data. In the application of Image processing,CRFs have the unique advantages over large scale spatial dependency, semantic informationand directly modelling posteri probability of a given image. Thus, the framework is capable toreduce uncertainty and improve the precision in image labelling.
Owe to the advantages mentioned above of CRFs theory, for some problems existing inobject class image segmentation based on image labelling, the thesis develops the studies onlow and higher order CRFs model when considering multi-scale intrinsic properties andsemantic feature of image, the details and innovative studies are in the following:
For the purpose of better express spatial, even semantic information of image, it isproposed that Probability Latent Semantic Analysis（PLSA）technology has been integratedinto the framework of CRFs via Support Vector Machine(SVM) classifier in order to achievean improved precision of object class segmentation in multi-label case. Details are as follows:Firstly, semantic features were derived from a PLSA clustering procedure can be an input toSVM classifier. Then, an output in the form of probability value from the well-defined SVMclassifier was used to make the associative potential function, Potts function and featurecontrast function were used to define the interactive potential function, then a PLSA-CRFscan be formed. In the process, an efficient Expectation Maximization（EM） algorithm wasintroduced to better estimate the parameters of PLSA model in order to improve less efficientproblem using standard EM algorithm. Finally, piecewise training algorithm is used toestimate the parameters of the proposed model, and Loopy Belief Propagation （LBP）algorithm is used to approximately infer the model.
For the problem “superpixel-based CRFs using features only from a scale can’t expressinherent multi-level spatial structure and semantic relation of an object”, the thesis develops apairwise superpixel-based CRFs model with association term defined as an output of RandomForest classifier based on multi-level spatial context and semantic features of objects so as toacquire a high precision in object class image segmentation. Main studies include: multi-levelspatial context feature extraction and boosted feature selection method, the definition ofassociative potential using Random Forest, and interaction potential weighted by common boundary of neighbors for CRFs model. Finally, pseudo-maximum likelihood method wasused to train the model for parameters estimation, and LBP algorithm was used to infer thepiecewise model.
For the problem of “the pixel-based CRFs is low efficient”,“the scale and quality ofimage over-segmentation have greater effects on superpixel-based CRFs, which have limitedability of spatial interaction” for image labeling, the thesis carried out the studies and makesimprovements on higher order CRFs model based on label consistency soft constraint andquality of image over-segmentation proposed by Kohli. Details are as follows: the model waswell studied by experiments on complex scenes at the first. A new segmentation qualitysensitive function for the purpose of image labelling (not image over-segmentation) was thenproposed, thus higher order potential was updated. Finally, the improved HoCRF in the thesisis trained by piecewise learning algorithm and inferred by transformed α-expansion based ongraph cut method.
Experiments on images including natural, aerial imagery with complex scene are carriedto verify the three proposed models and some relative methods. The results have shown thosemethods are effective to object class segmentation in multi-label case for those images such ascomplex natural scene and aerial images.
引文
[1]钟平.面向图像标记的随机场模型研究[D].湖南:国防科学技术大学,2008
[2]谢昭.图像理解的关键问题和方法研究[D].安徽：合肥工业大学，2007
[3]朱蓉.基于语义信息的图像理解关键问题研究[J].计算机应用研究，2009，26（4）:1237-1240
[4]高隽，谢昭，张骏，等.图像语义分析与理解综述[J],模式识别与人工智能，2010，23（2）：191-202
[5] Wang X, Zhang X P, Clarke I, et al. A New Image Labeling Method Based onContent-based Image Retrieval and Conditional Random Field[C]. InternationalSymposium on IEEE Image and Signal Processing and Analysis（ISPA）,IEEE,2009:221-226
[6]李玲玲，金泰松，李翠华.基于局部特征和隐条件随机场的场景分类方法[J].北京理工大学学报,2012,32(7):720-724.
[7] Chan T.F., Vese L.A. Active Contours Without Edges[J]. IEEE Transaction on ImageProcessing,2001,10(2):266-277
[8] Mansouri A.R., Mitiche A., Va′zquez C. Multiregion Competition: A Level Set Extensionof Region Competition to Multiple Region Image Partitioning[J].Computer Vision andImage Understanding,2006,101(3):137-150
[9] Chen S. Y., Tong H., Cattani C. Markov Models for Image Labeling[J]. MathematicalProblems in Engineering,2012. Mathematical Problems in Engineering,2012,2012(814356):18pages
[10]Lafferty J., McCallum A., Pereira F. Conditional Random Fields: Probabilistic Models forSegmenting and Labeling Sequence Data[C]. International Conference on MachineLearning, IEEE,2001:282-289
[11]Kumar S., Hebert M. Discriminative Random Fields: A Discriminative Framework forContextual Interaction in Cassification[C]. IEEE International Conference on ComputerVision, IEEE,2003:1150-l157
[12]Antunes A. F. B., Lingnau C., Da Silva J. C. Object Oriented Analysis and SemanticNetwork for High Resolution Image Classification[J]. Anais XI SBSR, Belo Horizonte,Brasil,2003:05-10
[13]Geng X., Zhao J. Interactive Image Segmentation with Conditional RandomFields[C].IEEE International Conference on Natural Computation, IEEE,2008,2:96-101
[14]张微,汪西莉.基于超像素的条件随机场图像分类计算机应用[J].2012,32(5):1272－1275,1279
[15]徐振宇，孙玉梅.基于条件随机场的多标签图像分类[J].辽宁工业大学学报（自然科学版）,2012,32(4):223-226
[16]褚一平,张引,叶修梓,等.基于隐条件随机场的自适应视频分割算法[J].自动化学报,2007,33(12),1252-1258
[17]高琳,唐鹏,盛鹏,等.复杂场景下基于条件随机场的视觉目标跟踪[J].光学学报,2010,30(6):1721-1728
[18]Yang M. Y., F rstner W., Drauschke M. Hierarchical Conditional Random Field forMulti-Class Image Classification[C]. International Conference on Computer VisionTheory andApplications. IEEE,2010,2:464-469
[19]He X., Zemel R, Carreira-perpinan M. Multiscale Conditional Random Fields for ImageLabeling[C]. IEEE Computer Society Conference on Computer Vision and PatternRecognition, IEEE,2004,2: II-695-II-702
[20]Plath N., Toussaint M., Nakajima S. Multi-class Image Segmentation Using ConditionalRandom Fields and Global Classification[C].26th Annual International Conference onMachine Learning. ACM,2009:817-824
[21]Gu S., Zheng Y., Tomasi C., Extended Pairwise Potentials[C]. CVPR Workshop onInference in Graphical Models with Structured Potentials,IEEE,2011
[22]Baloch S., Cheng E., Fang T. Shape based Conditional Random Fields for SegmentingIntracranial Aneurysms[M].Image-Based Geometric Modeling and Mesh Generation.Springer Netherlands,2013:55-67
[23]Zhong P., Wang R. Modeling and Classifying Hyperspectral Imagery by CRFs WithSparse Higher Order Potentials[J]. IEEE Transactions on Geoscience and RemoteSensing,2011,49(2):688-705
[24]Wegner J.D., H nsch R., Thiele A., et al. Building Detection From One Orthophoto andHigh-Resolution InSAR Data Using Conditional Random Fields[J]. IEEE Journal ofselected topics in applied Earth Observations and Remote Sensing,2011,4(1):83-91
[25]Kipkemboi B., Enschede K. Detection of Built-up Areas in SAR and Aster Images UsingConditional Random Fields[D]. The Netherlands: The Faculty of Geo-informationScience and Earth Observation of the University of Twente,2012
[26]王明军,周俊,屠珺,等.基于条件随机场的大范围地形感知框架等[J].机器人,2010,32(3):326-333
[27]Nguyen T.V., Pham N., Tran T., et al.Higher Order Conditional Random Field forMulti-Label Interactive Image Segmentation.RIVF, IEEE,2012:1-4.
[28]Wegner J. D., Montoya-Zegarra J.A., Schindler K. A Higher-order CRF Model for RoadNetwork Extraction[J]. IEEE Conference on Computer Vision and Pattern Recognition(CVPR), IEEE,2013:1698-1705
[29]Kohli P., Kumar M, Torr P., P3&Beyond: Solving Energies with Higher OrderCliques[C]. IEEE International Conference on Computer Vision and Pattern Recognition(CVPR), IEEE Computer Society Press,2007:1-8
[30]Kohli P., Ladicky L., Philip H. S. Torr, Robust Higher Order Potentials for EnforcingLabel Consistency[J]. International Journal of Computer Vision,2009,82(3):302-324
[31]Gonfaus J. M., Boix X., Van De Weijer J., et al. Harmony potentials for jointclassification and segmentation[C]. IEEE Conference on Computer Vision and PatternRecognition, IEEE,2010:3280-3287
[32]Boix X., Gonfaus J. M., van de Weijer J., et al. Harmony potentials[J]. InternationalJournal of Computer Vision, IEEE,2012,96(1):83-102
[33]Ladicky L., Russell C., Kohli P., et al. Associative Hierarchical CRFs for Object ClassImage Segmentation[C]. IEEE International Conference on Computer Vision, IEEE,2009:739-746
[34]Yang M.Y., F rstner W. A Hierarchical Conditional Random Field Model for Labelingand Classifying Images of Man-made Scenes[C]. IEEE International Conference onComputer Vision Workshops (ICCV Workshops), IEEE,2011:196-203
[35]Batra D., Sukthankar R., Chen T. Learning Class-Specific Affinities for ImageLabelling[C]. Proceedings of IEEE Conference on Computer Vision and PatternRecognition, IEEE,2008:1-8
[36]Russell C. Higher-Order Inference for Vision Problems[D]. Britain:University of London,2012
[37]史利民,余淼,胡占义.计算机视觉中高阶能量项的优化[J].计算机辅助设计与图形学学报,2011,23(5):731-734
[38]Kumar S., Hebea M. Discriminative random fields[J]. International Journal of ComputerVision,2006,68(2):179-201
[39]郭剑毅，薛征山，余正涛，等，基于层叠条件随机场的旅游领域命名实体识别[J].中文信息学报，2009,23(5):47-52
[40]Whittaker J. Graphical Models in Applied Multivariate Statistics[M]. Wiley Publishing,2009
[41]Koller D., Friedman N. Probabilistic Graphical Models: Principles and Techniques[M].Publisher: The MIT Press,2009, Pages:1208
[42]Li S. Z. Markov Random Field Models in Computer Vision[M]. Springer-Verlag NewYork, Inc.,1995
[43]Ngan K. N., Li H.L. Video Segmentation and Its Applications[M]. Springer,2011, XIV,164p.
[44]Galleguillos C., Belongie S. Context based Object Categorization: A Critical Survey[J].Computer Vision Image Understanding,2010,114(6):712-722
[45]Delong A., Osokin A., Isack H. N., et al. Fast Approximate Energy Minimization withLabel Costs[J]. International Journal of Computer Vision,2012,96(1):1-27
[46]Rother C., Nowozin S. Higher-order models in Computer Vision[EB/OL]. MachineLearning and Perception Group, Microsoft ResearchCambridge, UK,2010,http://research.microsoft.com/en-us/um/cambridge/projects/tutorials/cvpr10/
[47]Huang Q.X., Han M., Wu B., et al. A Hierarchical Conditional Random Field Model forLabeling and Images of Street Scenes[C]. International Conference on Computer Visionand Pattern Recognition,2011,1953-1960
[48]Lee C. H., Wang S., Jiao F., et al. Learning to Model Spatial Dependency:Semi-supervised Discriminative Random Fields [C].Advances in Neural InformationProcessing Systems,2006:793-800
[49]Lee C. H., Schmidt M., Murtha A., et al. Segmenting Brain Tumors with ConditionalRandom Fields and Support Vector Machines[M]. Computer Vision for BiomedicalImageApplications. Springer Berlin Heidelberg,2005:469-478
[50]Yuan J., Li J., Zhang B. Exploiting Spatial Context Constraints for Automatic ImageRegion Annotation[C]. International conference on Multimedia, ACM,2007:595-604
[51]Lafferty J., Zhu X., Liu Y. Kernel Conditional Random Fields: Representation and CliqueSelection[C]. International conference on Machine learning, ACM,2004:64
[52]Torralba A., Murphy K.P., Freeman W.T. Contextual Models for Object Detection UsingBoosted Random Fields[J]. Neural Information Processing Systems,2004:1401-1408
[53]Jiang W., Chang S.F., Loui A.C. Context-based Concept Fusion with Boosted ConditionalRandom Fields[C]. IEEE International Conference on Acoustics,Speech and SignalProcessing, IEEE,2007:949-952
[54]Shotton J., Winn J., Rother C., et al. TextonBoost for Image Understanding: Multi-ClassObject Recognition and Segmentation by Jointly Modeling Texture, Layout, and Context[J]. International Journal of Computer Vision,2009,81(1):2-23
[55]Kr henbühl P., Koltun V. Efficient Inference in Fully Connected CRFs with GaussianEdge Potentials[J]. arXiv preprint arXiv:1210.5644,2012
[56]Galleguillos C., Rabinovich A., Belongie S. Object Categorization Using Co-occurrence,Location and Appearance[C]. International Conference on Computer Vision and PatternRecognition, IEEE,2008:1-8
[57]Gould S., Rodgers J., Cohen D., et al. Multi-class Segmentation with Relative LocationPrior[J]. International Journal of Computer Vision,2008,80(3):300-316
[58]Fulkerson B., Vedaldi A., Soatto S. Class Segmentation and Object Localization withSuperpixel Neighborhoods[C]. International Conference on Computer Vision, IEEE,2009:670-677
[59]Nowozin S., Gehler P. V., Lampert C. H. On Parameter Learning in CRF-basedApproaches to Object Class Image Segmentation[M]. Computer Vision–ECCV2010.Springer Berlin Heidelberg,2010:98-111
[60]Roller B. Max-margin Markov Networks[J]. Proc. Advances in Neural InformationProcessing Systems, MIT Press,2004:25
[61]Munoz D., Bagnell J.A., Vandapel N., et al. Contextual Classification with FunctionalMax-margin Markov Networks[C]. International Conference on Computer Vision andPattern Recognition, IEEE,2009:975-982
[62]Kumar S., August J., Hebert M., Exploiting Inference for Approximate ParameterLearning in Discriminative Fields: An Empirical Study[C]. Energy MinimizationMethods in Computer Vision and Pattern Recognition. Springer Berlin Heidelberg,2005:153-168
[63]He X., Zemel R. S., Ray D. Learning and Incorporating Top-Down Cues in ImageSegmentation[M]. Computer Vision–ECCV2006. Springer Berlin Heidelberg,2006:338-351
[64]Kor F., F rstner W. Approximate Parameter Learning in Conditional Random Fields: AnEmpirical Investigation[M]. Pattern Recognition. Springer Berlin Heidelberg,2008:11-20
[65]Gould S. Max-margin Learning for Lower Linear Envelope Potentials in Binary MarkovRandom Fields[C]. International Conference on Machine Learning(ICML),2011:193-200
[66]Park K., Gould S. On Learning Higher-Order Consistency Potentials for Multi-class PixelLabeling[M]. Computer Vision–ECCV2012. Springer Berlin Heidelberg,2012:202-215
[67]Komodakis N., Paragios N., Tziritas G. MRF Energy Minimization and Beyond via DualDecomposition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(3):531-552
[68]Murphy K. P., Weiss Y., Jordan M.I. Loopy Belief Propagation forApproximate Inference:An Empirical Study[C]. International conference on Uncertainty in artificial intelligence.Morgan Kaufmann Publishers Inc.,1999:467-475
[69]Szeliski R., Zabih R., Scharstein D., et al. A Comparative Study of Energy MinimizationMethods for Markov Random Fields[M] Computer Vision–ECCV2006. Springer BerlinHeidelberg,2006:16-29
[70]Kappes J. H., Andres B., Hamprecht F. A., et al. A Comparative Study of ModernInference Techniques for Discrete Energy Minimization Problems[C]. CVPR,2013
[71]Russell C., Kohli P., Torr P. H. S. Exact and Approximate Inference in AssociativeHierarchical Networks Using Graph Cuts[J]. arXiv preprint arXiv:1203.3512,2012
[72]Hofmann T. Probabilistic Latent Semantic Indexing[C]. The22nd annual internationalACM SIGIR conference on Research and development in information retrieval, ACM,1999:50-57
[73]Hofmann T. Unsupervised Learning by Probabilistic Latent Semantic Analysis[J].Machine Learning,2001,42(1):177-196
[74]Blei D. M., Ng A. Y., Jordan M. I. Latent Dirichlet Allocation[J]. Journal of machineLearning research,2003,3:993-1022
[75]Bosch A., Zisserman A., Munoz X. Scene Classification via pLSA[M].ComputerVision–ECCV2006. Springer Berlin Heidelberg,2006:517-530
[76]Wang Y., Gong S.G. Conditional Random Field for Natural Scene Categorization[C].Britain Machine Vision Conference(BMVC), University of Warwick,2007:1-10
[77]陶超,谭毅华,彭碧发,等.一种基于概率潜在语义模型的高分辨率遥感图像分类方法[J].测绘学报,2011,40(2):156-162
[78]Wu X.Q., Amrikachi M., Shah S. K. Embedding Topic Discovery in Conditional RandomFields Model for Segmenting Nuclei Using Multispectral Data[J]. IEEE Transactions OnBiomedical Engineering,2012,59(6):1539-1549
[79]郑肇葆,郑宏.利用PLSA技术进行图像分割[J].武汉大学学报(信息科学版),2012,37(2):132-135
[80]Verbeek J., Triggs B. Region Classification with Markov Field Aspect Models[C].International Conference on Computer Vision&Pattern Recognition, IEEE,2007:1–8
[81]Yang W., Triggs B., Dai D.X., et al. Scene Segmentation via Low-dimensional SemanticRepresentation and CRFs[J]. EURASIP Journal on Advances in SignalProcessing,2010,2010(196036):14pages
[82]刘梦玲,何楚,苏鑫,等.基于pLSA和Topo-MRF模型的SAR图像分类算法研究[J].武汉大学学报(信息科学版),2011,36(1):122-125
[83]Castellani U., Perina A., Murino V., et al. Brain Morphometry by Probabilistic LatentSemantic Analysis[J]. Medical Image Computing and Computer-AssistedIntervention–MICCAI,2010:177-184
[84]Yang J., Jiang Y. G., Hauptmann A. G., et al. Evaluating Bag-of-visual-wordsRepresentations in Scene Classification[C]. Workshop on multimedia informationretrieval, ACM,2007:197-206
[85]Wang J., Liu P., She M. F. H., et al. Supervised Learning Probabilistic Latent SemanticAnalysis for Human MotionAnalysis[J]. Neurocomputing,2013,100:134-143
[86]Wu T. F., Lin C. J., Weng R. C. Probability Estimates for Multi-class Classification byPairwise Coupling[J]. Journal of Machine Learning Research,2004,5:975-1005
[87]Platt J. Probabilistic outputs for support vector machines and comparison to regularizedlikelihood methods [J].Advances in large margin classifiers,1999,10(3):61-74
[88]Hsu C. W., Lin C. J. A Comparison of Methods for Multiclass Support VectorMachines[J]. IEEE Transactions on Neural Networks,2002,13(2):415-425
[89]Maji S., Berg A C., Malik J. Classification Using Intersection Kernel Support VectorMachines is Efficient[C]. Conference on Computer Vision and PatternRecognition(CVPR), IEEE,2008:1-8
[90]Chang C.C., Lin C.J. LIBSVM: A Library for Support Vector Machines[CP/OL]. ACMTransactions on Intelligent Systems and Technology,2:27:1-27:27,2011, Softwareavailable at http://www.csie.ntu.edu.tw/~cjlin/libsvm
[91]Hsu C.,Chang C.C., W., Lin C. J.A Practical Guide to Support VectorClassification[CP/OL], http://www.csie.ntu.edu.tw/~cjlin,2010
[92]Sutton C., McCallum A. Piecewise Training for Undirected Models[J]. arXiv preprintarXiv:1207.1409,2012
[93]Andrew Y. N., Stephen G., Morgan Q.STAIR: The STanford Artificial Intelligence RobotProject[CP/OL], Ashutosh Saxena and Eric Berger. In Learning Workshop,http://stair.stanford.edu/，Snowbird,2008
[94]Shi J., Malik J. Normalized Cuts and Image Segmentation[J]. IEEE Transactions onPatternAnalysis and Machine Intelligence,2000,22(8):888-905
[95]Stefan K., Michael D., Horst B., Super-pixel Class Segmentation in Large-Scale AerialImagery[C]. Annual Workshop of theAustrian Association for Pattern Recognition,2010
[96]Zhang W.,Wang X. L. Superpixel-based Conditional Random Field for ImageClassification[J]. Journal of ComputerApplications,2012,32(5):1272-1275,1279
[97]Hao Z., Wang Q., Ren H., et al. Multiscale Superpixel Classification for TumorSegmentation in Breast Ultrasound Images[C]. International Conference on ImageProcessing (ICIP), IEEE,2012:2817-2820
[98]Shu G., Dehghan A., Shah M. Improving an Object Detector and Extracting Regionsusing Superpixels[J].2013
[99]Su X., He C., Feng Q., Deng X., et al. A Supervised Classification Method Based onConditional Random Fields With Multiscale Region Connection Calculus Model for SARImage[J]. IEEE Geoscience and Remote Sensing Letters,2011,8(3):497-501
[100] Breiman L. Random forests[J]. Machine learning,2001,45(1):5-32
[101] Statnikov A., Wang L., Aliferis C. F. A Comprehensive Comparison of RandomForests and Support Vector Machines for Microarray-based Cancer Classification[J].BMC bioinformatics,2008,9(1):319
[102] Bruzzone L., Carlin L. A Multilevel Context-based System for Classification of VeryHigh Spatial Resolution Images[J]. IEEE Transactions on Geoscience and RemoteSensing,2006,44(9):2587-2600
[103] Payet N., Todorovic S.(RF)^2--Random Forest Random Field[C].Advances inNeural Information Processing Systems.2010:1885-1893
[104] Comaniciu D., Meer P. Mean shift: A Robust Approach toward Feature SpaceAnalysis[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2002,24(5):603-619
[105] Felzenszwalb P. F., Huttenlocher D. P. Efficient Graph-based Image Segmentation[J].International Journal of Computer Vision,2004,59(2):167-181
[106] Jaimes A., Chang S. Model-Based Classification of Visual Information forContent-Based Retrieval[C]. Electronic Imaging'99. International Society for Optics andPhotonics,1998:402-414
[107]李德仁,童庆禧,李荣兴,等.高分辨率对地观测的若干前沿科学问题[J].中国科学:地球科学,2012,42(6):805-813
[108] Freund Y., Boosting A Weak Learning Algorithm by Majority[J]. Information andComputation,1995,121(2):256–285
[109] Kontschieder P., Bulo S. R., Bischof H., et al. Structured Class-labels in RandomForests for Semantic Image Labelling[C]. International Conference on Computer Vision(ICCV), IEEE,2011:2190-2197
[110]刘毅,杜培军,郑辉,等.基于随机森林的国产小卫星遥感影像分类研究[J].测绘科学,2012,37(4):194-196
[111] Rabinovich A., Vedaldi A., Galleguillos C., et al. Objects in Context[C]. InternationalConference on Computer Vision, IEEE,2007:1-8
[112] Gould S. DARWIN: A Framework for Machine Learning and Computer VisionResearch and Development[J]. Journal of Machine Learning Research (JMLR),2012:35333537
[113] Veksler O., Boykov Y., Mehrani P. Superpixels and Supervoxels in An EnergyOptimization Framework[M].Computer Vision–ECCV2010. Springer Berlin Heidelberg,2010:211-224
[114] Wang C., Paragios N., Markov Random Fields in Vision Perception: A Survey[R]France: Center for Visual Computing, école Centrale Paris, Chatenay-Malabry,2012
[115] Schoenemann T. Minimizing Count-based High Order Terms in Markov RandomFields[C]. Energy Minimization Methods in Computer Vision and PatternRecognition,Springer Berlin Heidelberg,2011:17-30
[116] Ramalingam S., Kohli P., Alahari K., et al. Exact Inference in Multi-label CRFs withHigher Order Cliques[C]. IEEE Conference on Computer Vision and Pattern Recognition,IEEE,2008:1-8
[2]谢昭.图像理解的关键问题和方法研究[D].安徽：合肥工业大学，2007
[3]朱蓉.基于语义信息的图像理解关键问题研究[J].计算机应用研究，2009，26（4）:1237-1240
[4]高隽，谢昭，张骏，等.图像语义分析与理解综述[J],模式识别与人工智能，2010，23（2）：191-202
[5] Wang X, Zhang X P, Clarke I, et al. A New Image Labeling Method Based onContent-based Image Retrieval and Conditional Random Field[C]. InternationalSymposium on IEEE Image and Signal Processing and Analysis（ISPA）,IEEE,2009:221-226
[6]李玲玲，金泰松，李翠华.基于局部特征和隐条件随机场的场景分类方法[J].北京理工大学学报,2012,32(7):720-724.
[7] Chan T.F., Vese L.A. Active Contours Without Edges[J]. IEEE Transaction on ImageProcessing,2001,10(2):266-277
[8] Mansouri A.R., Mitiche A., Va′zquez C. Multiregion Competition: A Level Set Extensionof Region Competition to Multiple Region Image Partitioning[J].Computer Vision andImage Understanding,2006,101(3):137-150
[9] Chen S. Y., Tong H., Cattani C. Markov Models for Image Labeling[J]. MathematicalProblems in Engineering,2012. Mathematical Problems in Engineering,2012,2012(814356):18pages
[10]Lafferty J., McCallum A., Pereira F. Conditional Random Fields: Probabilistic Models forSegmenting and Labeling Sequence Data[C]. International Conference on MachineLearning, IEEE,2001:282-289
[11]Kumar S., Hebert M. Discriminative Random Fields: A Discriminative Framework forContextual Interaction in Cassification[C]. IEEE International Conference on ComputerVision, IEEE,2003:1150-l157
[12]Antunes A. F. B., Lingnau C., Da Silva J. C. Object Oriented Analysis and SemanticNetwork for High Resolution Image Classification[J]. Anais XI SBSR, Belo Horizonte,Brasil,2003:05-10
[13]Geng X., Zhao J. Interactive Image Segmentation with Conditional RandomFields[C].IEEE International Conference on Natural Computation, IEEE,2008,2:96-101
[14]张微,汪西莉.基于超像素的条件随机场图像分类计算机应用[J].2012,32(5):1272－1275,1279
[15]徐振宇，孙玉梅.基于条件随机场的多标签图像分类[J].辽宁工业大学学报（自然科学版）,2012,32(4):223-226
[16]褚一平,张引,叶修梓,等.基于隐条件随机场的自适应视频分割算法[J].自动化学报,2007,33(12),1252-1258
[17]高琳,唐鹏,盛鹏,等.复杂场景下基于条件随机场的视觉目标跟踪[J].光学学报,2010,30(6):1721-1728
[18]Yang M. Y., F rstner W., Drauschke M. Hierarchical Conditional Random Field forMulti-Class Image Classification[C]. International Conference on Computer VisionTheory andApplications. IEEE,2010,2:464-469
[19]He X., Zemel R, Carreira-perpinan M. Multiscale Conditional Random Fields for ImageLabeling[C]. IEEE Computer Society Conference on Computer Vision and PatternRecognition, IEEE,2004,2: II-695-II-702
[20]Plath N., Toussaint M., Nakajima S. Multi-class Image Segmentation Using ConditionalRandom Fields and Global Classification[C].26th Annual International Conference onMachine Learning. ACM,2009:817-824
[21]Gu S., Zheng Y., Tomasi C., Extended Pairwise Potentials[C]. CVPR Workshop onInference in Graphical Models with Structured Potentials,IEEE,2011
[22]Baloch S., Cheng E., Fang T. Shape based Conditional Random Fields for SegmentingIntracranial Aneurysms[M].Image-Based Geometric Modeling and Mesh Generation.Springer Netherlands,2013:55-67
[23]Zhong P., Wang R. Modeling and Classifying Hyperspectral Imagery by CRFs WithSparse Higher Order Potentials[J]. IEEE Transactions on Geoscience and RemoteSensing,2011,49(2):688-705
[24]Wegner J.D., H nsch R., Thiele A., et al. Building Detection From One Orthophoto andHigh-Resolution InSAR Data Using Conditional Random Fields[J]. IEEE Journal ofselected topics in applied Earth Observations and Remote Sensing,2011,4(1):83-91
[25]Kipkemboi B., Enschede K. Detection of Built-up Areas in SAR and Aster Images UsingConditional Random Fields[D]. The Netherlands: The Faculty of Geo-informationScience and Earth Observation of the University of Twente,2012
[26]王明军,周俊,屠珺,等.基于条件随机场的大范围地形感知框架等[J].机器人,2010,32(3):326-333
[27]Nguyen T.V., Pham N., Tran T., et al.Higher Order Conditional Random Field forMulti-Label Interactive Image Segmentation.RIVF, IEEE,2012:1-4.
[28]Wegner J. D., Montoya-Zegarra J.A., Schindler K. A Higher-order CRF Model for RoadNetwork Extraction[J]. IEEE Conference on Computer Vision and Pattern Recognition(CVPR), IEEE,2013:1698-1705
[29]Kohli P., Kumar M, Torr P., P3&Beyond: Solving Energies with Higher OrderCliques[C]. IEEE International Conference on Computer Vision and Pattern Recognition(CVPR), IEEE Computer Society Press,2007:1-8
[30]Kohli P., Ladicky L., Philip H. S. Torr, Robust Higher Order Potentials for EnforcingLabel Consistency[J]. International Journal of Computer Vision,2009,82(3):302-324
[31]Gonfaus J. M., Boix X., Van De Weijer J., et al. Harmony potentials for jointclassification and segmentation[C]. IEEE Conference on Computer Vision and PatternRecognition, IEEE,2010:3280-3287
[32]Boix X., Gonfaus J. M., van de Weijer J., et al. Harmony potentials[J]. InternationalJournal of Computer Vision, IEEE,2012,96(1):83-102
[33]Ladicky L., Russell C., Kohli P., et al. Associative Hierarchical CRFs for Object ClassImage Segmentation[C]. IEEE International Conference on Computer Vision, IEEE,2009:739-746
[34]Yang M.Y., F rstner W. A Hierarchical Conditional Random Field Model for Labelingand Classifying Images of Man-made Scenes[C]. IEEE International Conference onComputer Vision Workshops (ICCV Workshops), IEEE,2011:196-203
[35]Batra D., Sukthankar R., Chen T. Learning Class-Specific Affinities for ImageLabelling[C]. Proceedings of IEEE Conference on Computer Vision and PatternRecognition, IEEE,2008:1-8
[36]Russell C. Higher-Order Inference for Vision Problems[D]. Britain:University of London,2012
[37]史利民,余淼,胡占义.计算机视觉中高阶能量项的优化[J].计算机辅助设计与图形学学报,2011,23(5):731-734
[38]Kumar S., Hebea M. Discriminative random fields[J]. International Journal of ComputerVision,2006,68(2):179-201
[39]郭剑毅，薛征山，余正涛，等，基于层叠条件随机场的旅游领域命名实体识别[J].中文信息学报，2009,23(5):47-52
[40]Whittaker J. Graphical Models in Applied Multivariate Statistics[M]. Wiley Publishing,2009
[41]Koller D., Friedman N. Probabilistic Graphical Models: Principles and Techniques[M].Publisher: The MIT Press,2009, Pages:1208
[42]Li S. Z. Markov Random Field Models in Computer Vision[M]. Springer-Verlag NewYork, Inc.,1995
[43]Ngan K. N., Li H.L. Video Segmentation and Its Applications[M]. Springer,2011, XIV,164p.
[44]Galleguillos C., Belongie S. Context based Object Categorization: A Critical Survey[J].Computer Vision Image Understanding,2010,114(6):712-722
[45]Delong A., Osokin A., Isack H. N., et al. Fast Approximate Energy Minimization withLabel Costs[J]. International Journal of Computer Vision,2012,96(1):1-27
[46]Rother C., Nowozin S. Higher-order models in Computer Vision[EB/OL]. MachineLearning and Perception Group, Microsoft ResearchCambridge, UK,2010,http://research.microsoft.com/en-us/um/cambridge/projects/tutorials/cvpr10/
[47]Huang Q.X., Han M., Wu B., et al. A Hierarchical Conditional Random Field Model forLabeling and Images of Street Scenes[C]. International Conference on Computer Visionand Pattern Recognition,2011,1953-1960
[48]Lee C. H., Wang S., Jiao F., et al. Learning to Model Spatial Dependency:Semi-supervised Discriminative Random Fields [C].Advances in Neural InformationProcessing Systems,2006:793-800
[49]Lee C. H., Schmidt M., Murtha A., et al. Segmenting Brain Tumors with ConditionalRandom Fields and Support Vector Machines[M]. Computer Vision for BiomedicalImageApplications. Springer Berlin Heidelberg,2005:469-478
[50]Yuan J., Li J., Zhang B. Exploiting Spatial Context Constraints for Automatic ImageRegion Annotation[C]. International conference on Multimedia, ACM,2007:595-604
[51]Lafferty J., Zhu X., Liu Y. Kernel Conditional Random Fields: Representation and CliqueSelection[C]. International conference on Machine learning, ACM,2004:64
[52]Torralba A., Murphy K.P., Freeman W.T. Contextual Models for Object Detection UsingBoosted Random Fields[J]. Neural Information Processing Systems,2004:1401-1408
[53]Jiang W., Chang S.F., Loui A.C. Context-based Concept Fusion with Boosted ConditionalRandom Fields[C]. IEEE International Conference on Acoustics,Speech and SignalProcessing, IEEE,2007:949-952
[54]Shotton J., Winn J., Rother C., et al. TextonBoost for Image Understanding: Multi-ClassObject Recognition and Segmentation by Jointly Modeling Texture, Layout, and Context[J]. International Journal of Computer Vision,2009,81(1):2-23
[55]Kr henbühl P., Koltun V. Efficient Inference in Fully Connected CRFs with GaussianEdge Potentials[J]. arXiv preprint arXiv:1210.5644,2012
[56]Galleguillos C., Rabinovich A., Belongie S. Object Categorization Using Co-occurrence,Location and Appearance[C]. International Conference on Computer Vision and PatternRecognition, IEEE,2008:1-8
[57]Gould S., Rodgers J., Cohen D., et al. Multi-class Segmentation with Relative LocationPrior[J]. International Journal of Computer Vision,2008,80(3):300-316
[58]Fulkerson B., Vedaldi A., Soatto S. Class Segmentation and Object Localization withSuperpixel Neighborhoods[C]. International Conference on Computer Vision, IEEE,2009:670-677
[59]Nowozin S., Gehler P. V., Lampert C. H. On Parameter Learning in CRF-basedApproaches to Object Class Image Segmentation[M]. Computer Vision–ECCV2010.Springer Berlin Heidelberg,2010:98-111
[60]Roller B. Max-margin Markov Networks[J]. Proc. Advances in Neural InformationProcessing Systems, MIT Press,2004:25
[61]Munoz D., Bagnell J.A., Vandapel N., et al. Contextual Classification with FunctionalMax-margin Markov Networks[C]. International Conference on Computer Vision andPattern Recognition, IEEE,2009:975-982
[62]Kumar S., August J., Hebert M., Exploiting Inference for Approximate ParameterLearning in Discriminative Fields: An Empirical Study[C]. Energy MinimizationMethods in Computer Vision and Pattern Recognition. Springer Berlin Heidelberg,2005:153-168
[63]He X., Zemel R. S., Ray D. Learning and Incorporating Top-Down Cues in ImageSegmentation[M]. Computer Vision–ECCV2006. Springer Berlin Heidelberg,2006:338-351
[64]Kor F., F rstner W. Approximate Parameter Learning in Conditional Random Fields: AnEmpirical Investigation[M]. Pattern Recognition. Springer Berlin Heidelberg,2008:11-20
[65]Gould S. Max-margin Learning for Lower Linear Envelope Potentials in Binary MarkovRandom Fields[C]. International Conference on Machine Learning(ICML),2011:193-200
[66]Park K., Gould S. On Learning Higher-Order Consistency Potentials for Multi-class PixelLabeling[M]. Computer Vision–ECCV2012. Springer Berlin Heidelberg,2012:202-215
[67]Komodakis N., Paragios N., Tziritas G. MRF Energy Minimization and Beyond via DualDecomposition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(3):531-552
[68]Murphy K. P., Weiss Y., Jordan M.I. Loopy Belief Propagation forApproximate Inference:An Empirical Study[C]. International conference on Uncertainty in artificial intelligence.Morgan Kaufmann Publishers Inc.,1999:467-475
[69]Szeliski R., Zabih R., Scharstein D., et al. A Comparative Study of Energy MinimizationMethods for Markov Random Fields[M] Computer Vision–ECCV2006. Springer BerlinHeidelberg,2006:16-29
[70]Kappes J. H., Andres B., Hamprecht F. A., et al. A Comparative Study of ModernInference Techniques for Discrete Energy Minimization Problems[C]. CVPR,2013
[71]Russell C., Kohli P., Torr P. H. S. Exact and Approximate Inference in AssociativeHierarchical Networks Using Graph Cuts[J]. arXiv preprint arXiv:1203.3512,2012
[72]Hofmann T. Probabilistic Latent Semantic Indexing[C]. The22nd annual internationalACM SIGIR conference on Research and development in information retrieval, ACM,1999:50-57
[73]Hofmann T. Unsupervised Learning by Probabilistic Latent Semantic Analysis[J].Machine Learning,2001,42(1):177-196
[74]Blei D. M., Ng A. Y., Jordan M. I. Latent Dirichlet Allocation[J]. Journal of machineLearning research,2003,3:993-1022
[75]Bosch A., Zisserman A., Munoz X. Scene Classification via pLSA[M].ComputerVision–ECCV2006. Springer Berlin Heidelberg,2006:517-530
[76]Wang Y., Gong S.G. Conditional Random Field for Natural Scene Categorization[C].Britain Machine Vision Conference(BMVC), University of Warwick,2007:1-10
[77]陶超,谭毅华,彭碧发,等.一种基于概率潜在语义模型的高分辨率遥感图像分类方法[J].测绘学报,2011,40(2):156-162
[78]Wu X.Q., Amrikachi M., Shah S. K. Embedding Topic Discovery in Conditional RandomFields Model for Segmenting Nuclei Using Multispectral Data[J]. IEEE Transactions OnBiomedical Engineering,2012,59(6):1539-1549
[79]郑肇葆,郑宏.利用PLSA技术进行图像分割[J].武汉大学学报(信息科学版),2012,37(2):132-135
[80]Verbeek J., Triggs B. Region Classification with Markov Field Aspect Models[C].International Conference on Computer Vision&Pattern Recognition, IEEE,2007:1–8
[81]Yang W., Triggs B., Dai D.X., et al. Scene Segmentation via Low-dimensional SemanticRepresentation and CRFs[J]. EURASIP Journal on Advances in SignalProcessing,2010,2010(196036):14pages
[82]刘梦玲,何楚,苏鑫,等.基于pLSA和Topo-MRF模型的SAR图像分类算法研究[J].武汉大学学报(信息科学版),2011,36(1):122-125
[83]Castellani U., Perina A., Murino V., et al. Brain Morphometry by Probabilistic LatentSemantic Analysis[J]. Medical Image Computing and Computer-AssistedIntervention–MICCAI,2010:177-184
[84]Yang J., Jiang Y. G., Hauptmann A. G., et al. Evaluating Bag-of-visual-wordsRepresentations in Scene Classification[C]. Workshop on multimedia informationretrieval, ACM,2007:197-206
[85]Wang J., Liu P., She M. F. H., et al. Supervised Learning Probabilistic Latent SemanticAnalysis for Human MotionAnalysis[J]. Neurocomputing,2013,100:134-143
[86]Wu T. F., Lin C. J., Weng R. C. Probability Estimates for Multi-class Classification byPairwise Coupling[J]. Journal of Machine Learning Research,2004,5:975-1005
[87]Platt J. Probabilistic outputs for support vector machines and comparison to regularizedlikelihood methods [J].Advances in large margin classifiers,1999,10(3):61-74
[88]Hsu C. W., Lin C. J. A Comparison of Methods for Multiclass Support VectorMachines[J]. IEEE Transactions on Neural Networks,2002,13(2):415-425
[89]Maji S., Berg A C., Malik J. Classification Using Intersection Kernel Support VectorMachines is Efficient[C]. Conference on Computer Vision and PatternRecognition(CVPR), IEEE,2008:1-8
[90]Chang C.C., Lin C.J. LIBSVM: A Library for Support Vector Machines[CP/OL]. ACMTransactions on Intelligent Systems and Technology,2:27:1-27:27,2011, Softwareavailable at http://www.csie.ntu.edu.tw/~cjlin/libsvm
[91]Hsu C.,Chang C.C., W., Lin C. J.A Practical Guide to Support VectorClassification[CP/OL], http://www.csie.ntu.edu.tw/~cjlin,2010
[92]Sutton C., McCallum A. Piecewise Training for Undirected Models[J]. arXiv preprintarXiv:1207.1409,2012
[93]Andrew Y. N., Stephen G., Morgan Q.STAIR: The STanford Artificial Intelligence RobotProject[CP/OL], Ashutosh Saxena and Eric Berger. In Learning Workshop,http://stair.stanford.edu/，Snowbird,2008
[94]Shi J., Malik J. Normalized Cuts and Image Segmentation[J]. IEEE Transactions onPatternAnalysis and Machine Intelligence,2000,22(8):888-905
[95]Stefan K., Michael D., Horst B., Super-pixel Class Segmentation in Large-Scale AerialImagery[C]. Annual Workshop of theAustrian Association for Pattern Recognition,2010
[96]Zhang W.,Wang X. L. Superpixel-based Conditional Random Field for ImageClassification[J]. Journal of ComputerApplications,2012,32(5):1272-1275,1279
[97]Hao Z., Wang Q., Ren H., et al. Multiscale Superpixel Classification for TumorSegmentation in Breast Ultrasound Images[C]. International Conference on ImageProcessing (ICIP), IEEE,2012:2817-2820
[98]Shu G., Dehghan A., Shah M. Improving an Object Detector and Extracting Regionsusing Superpixels[J].2013
[99]Su X., He C., Feng Q., Deng X., et al. A Supervised Classification Method Based onConditional Random Fields With Multiscale Region Connection Calculus Model for SARImage[J]. IEEE Geoscience and Remote Sensing Letters,2011,8(3):497-501
[100] Breiman L. Random forests[J]. Machine learning,2001,45(1):5-32
[101] Statnikov A., Wang L., Aliferis C. F. A Comprehensive Comparison of RandomForests and Support Vector Machines for Microarray-based Cancer Classification[J].BMC bioinformatics,2008,9(1):319
[102] Bruzzone L., Carlin L. A Multilevel Context-based System for Classification of VeryHigh Spatial Resolution Images[J]. IEEE Transactions on Geoscience and RemoteSensing,2006,44(9):2587-2600
[103] Payet N., Todorovic S.(RF)^2--Random Forest Random Field[C].Advances inNeural Information Processing Systems.2010:1885-1893
[104] Comaniciu D., Meer P. Mean shift: A Robust Approach toward Feature SpaceAnalysis[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2002,24(5):603-619
[105] Felzenszwalb P. F., Huttenlocher D. P. Efficient Graph-based Image Segmentation[J].International Journal of Computer Vision,2004,59(2):167-181
[106] Jaimes A., Chang S. Model-Based Classification of Visual Information forContent-Based Retrieval[C]. Electronic Imaging'99. International Society for Optics andPhotonics,1998:402-414
[107]李德仁,童庆禧,李荣兴,等.高分辨率对地观测的若干前沿科学问题[J].中国科学:地球科学,2012,42(6):805-813
[108] Freund Y., Boosting A Weak Learning Algorithm by Majority[J]. Information andComputation,1995,121(2):256–285
[109] Kontschieder P., Bulo S. R., Bischof H., et al. Structured Class-labels in RandomForests for Semantic Image Labelling[C]. International Conference on Computer Vision(ICCV), IEEE,2011:2190-2197
[110]刘毅,杜培军,郑辉,等.基于随机森林的国产小卫星遥感影像分类研究[J].测绘科学,2012,37(4):194-196
[111] Rabinovich A., Vedaldi A., Galleguillos C., et al. Objects in Context[C]. InternationalConference on Computer Vision, IEEE,2007:1-8
[112] Gould S. DARWIN: A Framework for Machine Learning and Computer VisionResearch and Development[J]. Journal of Machine Learning Research (JMLR),2012:35333537
[113] Veksler O., Boykov Y., Mehrani P. Superpixels and Supervoxels in An EnergyOptimization Framework[M].Computer Vision–ECCV2010. Springer Berlin Heidelberg,2010:211-224
[114] Wang C., Paragios N., Markov Random Fields in Vision Perception: A Survey[R]France: Center for Visual Computing, école Centrale Paris, Chatenay-Malabry,2012
[115] Schoenemann T. Minimizing Count-based High Order Terms in Markov RandomFields[C]. Energy Minimization Methods in Computer Vision and PatternRecognition,Springer Berlin Heidelberg,2011:17-30
[116] Ramalingam S., Kohli P., Alahari K., et al. Exact Inference in Multi-label CRFs withHigher Order Cliques[C]. IEEE Conference on Computer Vision and Pattern Recognition,IEEE,2008:1-8