资源描述：
《automatic face recognition system by combining four individual algorithms》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。
1、AutomaticFaceRecognitionSystembyCombiningFourIndividualAlgorithmsSection1:INTRODUCTIONUniquefacialcharacteristicsofhumanbeingsoftenrecognizeoneanother.Inthefacerecognitionapproach,agivenfaceiscomparedwiththefacesstoredinafacedatabaseinordertoidentifytheperson.Thepurposeistofindaf
2、aceinthedatabase,whichhasthehighestsimilaritywiththegivenface.Amongthemethodswehaveachieved,theprojectingapproachesarethosebasedonPrincipalComponentAnalysis(PCA),TemplateMatchingusingCorrelation,NeuralNetwork,ModelMatching,PartitionedIteratedFunctionSystem(PIFS),WaveletandDiscret
3、eCosineTransform(DCT).Forenhancingtheaccuracyandtoincreasetheperformanceofthefacerecognitionsystemacombinationofthesealgorithmsinagroupoftwoorthreeindividualtechniquescanbeused.here,wediscussthefourdifferentindividualfacerecognitiontechniquesbasedonPCA,DCT,TemplateMatchingusingCo
4、rrelation,PIFSandalsodiscussrecentlydevelopeddifferentmulti-algorithmicapproachesforthefacerecognitionsystemsbasedonthedifferentcombinations(ingroupoftwoorthree)oftheseindividualtechniques,andweusethesamemulti-algorithmicapproachforthefacerecognitionsystemsbasedonthedifferentcomb
5、inationsbutingroupoffouroftheseindividualtechniques.Thegoalistofindwhichcombinationofthesetechniquesperformsbetterintermsoffacerecognitionrate.Section2:RELATEDWORKS1.PCA(principlecomponentsanalysis,PCA)PCAalsoknownasEigenfacemethod,InPCAmethodtheimagesareprojectedontothefacialval
6、uesocalledeigenspace.PCAapproachreducesthedimensionofthedatabymeansofbasicdatacompressionmethodandrevealsthemosteffectivelowdimensionalstructureoffacialpatterns.2.LFA(localfeatureanalysis,LFA)LFAmethodofrecognitionisbasedontheanalysisthefaceintermsoflocalfeaturese.g.eye,noseetc.b
7、ywhatisreferredLFAkernels.3.NeuralNetworkRecognitionbyNeuralNetworkandarebasedonlearningofthefacesinan‘ExampleSet’bythemachineinthe‘TrainingPhase’andcarryingoutrecognitioninthe‘GeneralizationPhase’.4.SVM(SupportVectorMachines,SVM)SupportVectorMachines(SVM)techniqueisinfactoneofth
8、ebinaryclassificationmethods.Thesupportv
