欢迎来到天天文库
浏览记录
ID:54366941
大小:922.16 KB
页数:7页
时间:2020-04-29
《应用可见-近红外高光谱分析烟叶的颜色和部位特征.pdf》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、第34卷,第10期光谱学与光谱分析Vol畅34,No畅10,pp2758‐27632014年10月SpectroscopyandSpectralAnalysisOctober,2014AnalysisofTobaccoColorandLocationFeaturesUsingVisible‐NearInfraredHyperspectralData1122CAIJia‐yue,LIANGMiao,WENYa‐dong,YUChun‐xia,2211倡WANGLuo‐ping,WANGYi,ZHAOLong‐lian,LIJun‐hui1.CollegeofInformatio
2、nandElectricalEngineering,ChinaAgriculturalUniversity,Beijing 100083,China2.TechnologyCenterofYunnanTobaccoIndustryCo.,Ltd.,Kunming 650231,ChinaAbstractInthepresentpaper,sixcategoriesofstandardindustrialgradingtobaccoprovidedbyHongtaGrouparetakenasexperimentalsamples,includingthreedifferent
3、tobaccolocations‐upper(B),middle(C)andlower(X)parts,witheachpartcontainingtwokindsoftobaccocolors‐orange(O)andlemonyellow(L).TwomethodsincludingprojectionmodelmethodbasedonprincipalcomponentandFishercriterion(PPF)andsupportvectormachine(SVM)methodareusedtoanalyzecolorandlocationfeaturesofto
4、baccobasedonvisi‐ble‐nearinfraredhyperspectraldata.Theresultsofprojectionmodelmethodindicatethatintheprojectionandsimilarityanalysisoftobaccocolor,locationandsixtobaccogroupsclassifiedbycolorandlocation,twokindsofcolorcanbefullydifferentiated,ofwhichthesimilarityvalueis-1畅0008.Tobaccofromup
5、perandlowerpartscanalsobefullydifferentiatedwithsimilarityvalue0畅4053,buttheybothhaveintersectionswithtobac‐cofrommiddlepart.Sixtobaccogroupsclassifiedbycolorandlocationcanbefullydifferentiatedaswellandtheirprojectionpositionsmeettheactualexternalfeaturesoftobacco.Theresultsofsupportvectorm
6、achinemethodindicatethatinthediscriminantanalysisoftobaccocolor,locationandsixtobaccogroupsclassifiedbycolorandlocation,theaveragerecognitionrateoftobaccocolorsreaches98%.Theaveragerecognitionrateoftobaccolocationis96%.Theaveragerecognitionrateofsixtobaccogroupsis94%.Therefore,it’sfeasiblet
7、oanalyzecolorandlocationfeaturesoftobaccousingvisible‐nearinfraredhyperspectraldata,whichcanprovidereferencefortobaccoqualityevaluation,computer‐aidedgradingandtobaccointelligentacquisition,andalsooffersanewapproachtotheanalysisofexteriorfeaturesofothera
此文档下载收益归作者所有