OALib Journal期刊
ISSN: 2333-9721
费用:99美元
|
|
|
基于sda和cart算法的面向对象分类研究
DOI: 10.3969/j.issn.1000-2006.2015.03.002, PP. 6-12
Keywords: 面向对象分类,逐步判别分析,cart模型,敏感度评价
Abstract:
以2005年深圳市福田区quickbird影像为主要数据源,根据面向对象多尺度分割结果,构建和分析了不同地类对象的光谱、形状和纹理信息特征,在此基础上利用逐步判别分析法(stepwisediscriminantanalysis,sda)结合分类回归树(classificationandregressiontrees,cart)构建多尺度、多变量分类模型。结果表明:①采用sda在一定程度上能够客观、准确地进行特征子集预筛选,笔者从32个特征中筛选出27个特征用于构建cart模型,并按其区分地类能力进行了重要性排序,其中光谱特征与纹理特征排序比较靠前,形状特征中仅lengthwidth排在第5位,剩余特征比较靠后。②利用分类回归树模型可进一步优化特征选取,并智能化计算出分离阈值,基本实现面向对象的自动化分类。其中mean_b3、lengthwidth、ratio_b3、gldve和ndvi是重要分类节点。③逐步判别分析结合分类回归树构建的分类模型,可以在提高或者不显著降低影像分类精度的条件下实现特征降维。当取相关指数r2>0.2时,构建的分类模型效果最优,cart16模型训练和验证精度分别为94.44%和83.37%,其特征子集规模最小,与原始特征数量相比减少了一半。
References
[1] | 郑雅敏.基于遗传算法的特征选择方法的改进研究[d].重庆:重庆大学,2008.zhengym.improvementoffeatureselectionmethodbasedongeneticalgorithm[d].chongqing:chongqinguniversity,2008.
|
[2] | kononenkoi.estimationattributes:analysisandextensionsofrelief[c]//machinelearning:the1994europeanconferenceonmachinelearning.catania,1994:174-182.
|
[3] | johngh,kohavir,pflegerk.irrelevantfeaturesandthesubsetselectionproblem[c]//machinelearning:procofthe11thintconfonmachinelearning.newbrunswick,1994:121-129.
|
[4] | 姚旭,王晓丹,张玉玺,等.特征选择方法综述[j].控制与决策,2012,27(2):161-166.yaox,wangxd,zhangyx,etal.summaryoffeatureselectionalgorithms[j].controlanddecision,2012,27(2):161-166.
|
[5] | 吴浩苗,尹中航,孙富春.relief算法在笔迹识别的应用[j].计算机应用,2006,26(1):174-177.wuhm,yizh,sunfc.applicationofreliefinhandwritingrecognition[j].journalofcomputerapplications,2006,26(1):174-177.
|
[6] | 张克军,刘哲.基于改进的fcm和kpca的多光谱图像特征提取方法[j].科学技术与工程,2007,7(8):1657-1661.zhangkj,liuz.featureextractionformultispectralimagesbasedonimprovedfcmandkpca[j].sciencetechnologyandengineering,2007,7(8):1657-1661.
|
[7] | 余晓敏,湛飞并,廖明生,等.利用改进seath算法的面向对象分类特征选择方法[j].武汉大学学报:信息科学版,2012,37(8):921-924.yuxm,zhanfb,liaoms,etal.object-orientedfeatureselectionalgorithmsbasedonimprovedseathalgorithms[j].geomaticsandinformationscienceofwuhanuniversity,2012,37(8):921-924.
|
[8] | ruiliangpu,shawnlandry.acomparativeanalysisofhighspatialresolutionikonosandworldview-2imageryformappingurbantreespecies[j].remotesensingofenvironment,2012,124:516-533.
|
[9] | 潘琛,林怡,陈映鹰.基于多特征的遥感影像决策树分类[j].光电子?激光,2010,21(5):731-736.panc,liny,chenyy.decisiontreeclassificationofremotesensingimagesbasedonmulti-feature[j].journalofoptoelectronicslaser,2010,21(5):731-736.
|
[10] | 唐银凤,黄志明,黄荣娟,等.基于多特征提取和svm分类器的纹理图像分类[j].计算机应用与软件,2011,28(6):22-25.tangyf,huangzm,huangrj,etal.textureimageclassificationbasedonmulti-featureextractionandsvmclassifier[j].computerapplicationsandsoftware,2011,28(6):22-25.
|
[11] | 吴芳,贾永红.基于pca―bpnn的多光谱遥感影像分类[j].地理空间信息,2006,4(1):15-17.wuf,jiayh.classificationofmultispectralremotesensingimagebasedonpca-bpnn[j].geospatialinformation,2006,4(1):15-17.
|
[12] | 吴红霞,吴悦,刘宗田,等.基于relief和svm-rfe的组合式snp特征选择[j].计算机应用研究,2012,29(6):2074-2077.wuhx,wuy,liuzt,etal.combinedsnpfeatureselectionbasedonreliefandsvm-rfe[j].applicationresearchofcomputer,2012,29(6):2074-2077.
|
[13] | 张颖娜,张超,赵冬玲,等.利用relieffo和海明遗传算法的面向对象分类特征选择算法[j].武汉大学学报:信息科学版,2010,35(12):1444-1448.zhangyn,zhangc,zhaodl,etal.object-orientedfeatureselectionalgorithmswithrelieffoandhaiminggeneticalgorithm[j].geomaticsandinformationscienceofwuhanuniversity,2010,35(12):1444-1448.
|
[14] | 潘洁,李明诗.基于信息量的高分辨率影像纹理提取的研究[j].南京林业大学学报:自然科学版,2010,34(4):129-134.panj,lims.texturalfeaturesanalysisofhigh-resolutionremotesensingimagebasedontheinformationabundance[j].journalofnanjingforestryuniversity:naturalsciencesedition,2010,34(4):129-134.
|
[15] | marthatr,kerlen,vanwestencj,etal.segmentoptimizationanddata-driventhresholdingforknowledge-basedlandslidedetectionbyobject-basedimageanalysis[j].ieeetransactionsongeoscienceandremotesensing,2011,49(12):4928-4943.
|
[16] | stumpfa,kerlen.object-orientedmappingoflandslidesusingrandomforests[j].remotesensingofenvironment,2011,115(10):2564-2577.
|
[17] | 王登峰,杨志刚,魏安世.纹理信息在遥感影像分类中的应用[j].南京林业大学学报:自然科学版,2010,34(3):97-100.wangdf,yangzg,weias.applicationoftextureinformationonclassificationofremotesensingimagery[j].journalofnanjingforestryuniversity:naturalsciencesedition,2010,34(3):97-100.
|
[18] | mladinichcs.anevaluationofobject-orientedimageanalysistechniquestoidentifymotorizedvehicleeffectsinsemi-aridtoaridecosystemsoftheamericanwest[j].gisscience&remotesensing,2010,47(1):53-77.
|
Full-Text
|
|
Contact Us
service@oalib.com QQ:3279437679 
WhatsApp +8615387084133
|
|