基于2D-PCA的人臉識別方法研究.pdf

1、東北師范大學(xué)碩士學(xué)位論文基于2DPCA的人臉識別方法研究姓名：林玲申請學(xué)位級別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：孔俊20080501AbstraCtWiththerapiddevelopmentofinfomationtechn0109y，inf0珊ationsecurit)rbecomesmoreandmoreimportanttodaySomeapplicationdomainsneede銜cientautomaticperson

2、alidentificationtecllIlolo臥suchasElec仃onCommerce，Elec仃onBalll(，NeMorkSecurityandsoonBiometricsgetsmoreaIldmorewidelyapplicationinViewofitss訕ili壩uniqueness鋤dconvenienceFromthewarrantyofen仃ancetolockcriminalincrowd，all羽嘭re

3、latedt0themarketof印plicationand缸endof如tIlreofthistechn0109yBiometricscombinestlleinfomationtecllIlologywinlbiologytechnology，whichuseshum鋤iIlllerentbiologicalchamcteristicssuchasface，palm塒nt，and試s，bchavioralcharacteristi

4、cssuchasgait，signatureandspeechtoconfimpersonalidentityf0rr叩1acingors仃engtheningthetraditionalpersonalidentificationapproachesFacerecogIlitionhasVeDrla瑪eacademicaIldpracticalValuesIndailylife，peopleknowingeachotherusesat

5、mostofperson’sfaceTheVisualinfomationrenectedbyfacehasimponantmeaningandimpactbetweenpeople’sintercommunionaIldintercourseBecauseofitsextensiVealldappliedrealm，facerecognitiontechniquehasgottheextensiVeconcemwithsmdyinne

6、arthreedecadesandbecomethemostpotentialmethodofidentityreeognitionThispaperproposedonekindofIIllproVedtwo—dimensionalprincipalcomponentsanalysiswhich12DPCA，thepersonalfacereco嘶tionmethods，onhaVingdonethebasisstudyingwith

7、thorou曲in釅edientanalysistomeprincipalcomponentfirstly’t0ca幣esonthepretreatnlenttothefaceimage，theuseofthe仃aillingsamplec01lection，the12nPCAmethodwhichpresentsinthispapercalculatetheprincipalcomponent，thedefinitioncharac—

8、teristicspace，a仃ainingsarnpleofeachcharacteristicoftllecoefficientmatrix；secondly，theresearChgotagroupofcharacteristiccoefficientmatrixofdetectingmatonthesamplewithCarIjesonthesimilarpre缸eatlnentoperationinmeverification

9、staget0nleexamina—tionsample，andmapsitabovethecharactedsticspacewhichcalculatesbased0nthetrainingsamplecomes；Illtheend，theeXperimentwereusingtheminimumfromclassificationimplementt0beinprogressforidentificationcalculating

10、distancebe咐∞ntkmoduluscheckingand仃ainingthesamplebookcharacteristicWbrespectiVelyinORLda訕aseYaleBdatabase，Ⅱ1eAMPda切base1lasca曬edoutaneXp舐mentVerified2DPCAfacereco印itionmethodsValiditywrhich“spaperproposed，theexperimentha

11、dindicatedthatthemethodofthisp印erproposedhasthegoodrecognitioneffectinthecalibrationdatabasewithcarr)ringonthecontmstbetweenthemetllodwhichpro—posedtothispaper鋤dtraditional2DPCAKeywords：F∽eRecogllition；PrincipalCompone鵬A