線性子空間法人臉識別的研究

1、山東大學碩士學位論文線性子空間法人臉識別的研究姓名：曲玉強申請學位級別：碩士專業(yè)：信號與信息處理指導教師：吳曉娟20081023山東大學碩士學位論文ABSTRACTFacerecognitionhasbecomeahotresearchedfieldinbiometricrecognitionfieldComparing、航tIlothertechnologiesofbiometricrecognitionthetechnologyof

2、facerecognitionhasSOmanyad餾Ita舀；ssuchasnonintrusiveeasycollectionandhighaccuraccythatitispopularTherearebroadapplicationsinthefieldsofcommercial，lawenforcementpublicsecurityidentityclassificationentrancecontrol，videosurv

3、eillanceandSOOILFacerecognitionincludesthreeparts：facedetectionandlocalizationfeatureextractionandclassificationThisthesispresentsthestudiesoffeatureextractioninfacerecognitionFeatureextractionisaprocesswhichtransfersthe

4、datafromprimaryspacesintofeaturespace，representingtheminalowerdimensionalspacewithlesseffectivecharactersuptonow，manymethodSoffeatureextractionhavebeenproposedAmongthem，thesubspaceanalysishasreceivedextensiveattentionowi

5、ngtoitsappealingpropertiesNow，thesubspaceanalysismethodhasbeenthemostpopulartechnologyforfeatureextractionandfacerecognitionFeatureextractionbased011linearsubspaceanalysishasadvantages，suchashi曲computingefficencyandstron

6、ggeometryfeaturedescriptionabilityEigenfacemethodbasedonprincipalcomponentanalysisandthemethodbasedonlineardiscriminantanalysisaretwosuceuessfulonesBecameofthedisturbancecomingfromenvironmentoneorderfeatureextractionbase

7、donPCAorLDACanhardlygetthefeaturesexactlystandingforidentificationBycomparingtheselinearsubspaceanalysismethodsintheoryandexperimentresults，Thisthesiswillpointoutthatlineardiscriminateanalysisismoresuitableforfacerecogni

8、tionthanothersubspacemethodsFisherlineardiscriminateanalysishasobtainedsuccessfulapplicationsinfacerecognitionAnewLDA—basedfacerecognitionalgorithmisproposedinthethesiswhichiscalled“weightednullspaceapproach’Duetothefact