支持向量机及其在纹理分类中的应用-support vector machine and its application in texture classification.docxVIP

AnimprovedgridsearchapproachispresentedforparameterselectionofSup-portVectorMachine.AnoverallapplicationframeworkforSupportVectorMachineisproposedbasedonthekernelselectionandparametersearchstrategy.ThispaperalsousesSupportVectorMachineintextureclassification.Theper-formanceinfluenceofdifferentkernelfunctionanddifferenttexturefeatureisanalyzed.TheclassificationresultsshowthatRBFkernelcangivethebestper-formanceinmostclassificationgroup.Italsoshowsthatautocorrelationkernelfunction,whichconstructedwithautocorrelationfunctionusedfrequentlyinimageprocessing,doesnotperformwell.WhenusingSupportVectorMachinefortextureclassification,highdimensiontexturefeaturesfromthesameseriescangivemorepromisingresults.AncomparativestudyisperformedonSupportVectorMachinewithartificialneuralnetworkandLVQclassifier.ThebetterandmorestableperformanceshowsthatSupportVectorMachineissuitablefortextureclassification.Keywords:TextureClassification;SupportVectorMachine;StatisticalLearningTheory;KernelFunction;ParameterSelection目录中文摘要IAbstractIII目录VII第1章概论11.1论题的背景及意义............................11.2支持向量机................................21.2.1研究历史及现状..........................21.2.2支持向量机的基本思想及优点..................41.2.3目前支持向量机研究的内容...................61.3纹理分类.................................71.3.1纹理定义.............................71.3.2纹理分析及应用..........................101.3.3纹理分类及难点..........................111.3.4为什么将支持向量机用于纹理分类...............121.4本文的工作................................131.5论文的组织................................14第2章支持向量机及其实现152.1支持向量机的理论基础..........................152.1.1学习过程一致性条件.......................152.1.2VC维和推广性边界162.1.3经验风险最小化和结构风险最小化...............17V2.2线性支持向量机(LSVM)192.2.1线性可分问题...........................192.2.2近似线性可分问题........................202.3非线性支持向量机............................212.4支持向量机的实现算法..........................232.4.1选块算法.............................232.4.2分解算法.............................242.4.3序贯最小最优化算法.......................252.5小结....................................26