特征点距离及同卵双胞胎指纹识别方向领域(IJIGSP-V5-N2-8)

I.J.Image,GraphicsandSignalProcessing,2013,2,51-59PublishedOnlineFebruary2013inMECS()DOI:10.5815/ijigsp.2013.02.08Copyright©2013MECSI.J.Image,GraphicsandSignalProcessing,2013,2,51-59MinutiaeDistancesandOrientationFieldsBasedThumbprintIdentificationofIdenticalTwinsKamtaNathMishra*DepartmentofCS&E,BirlaInstituteofTechnologyMesra,Ranchi(AllahabadCampus),INDIAEmail:[e-mail]&E,BirlaInstituteofTechnology,MesraRanchi(AllahabadCampus),INDIAEmail:[e-mail]ntofInformationTechnology,IndianInstituteofInformationTechnologyAllahabad,INDIAEmail:[e-mail].,Bangaluru,INDIAEmail:[e-mail]CS&E,BirlaInstituteofTechnology,MesraRanchi(AllahabadCampus),INDIAEmail:[e-mail]—Thetwinsareclassifiedintotwocategoriesnamelyfraternalandidenticaltwins.FraternaltwinsdifferinfacestructuresandDNAsequencesbut,identicaltwinshavethesamefacestructureandsharesameDNAsequence.Therefore,itisdifficulttoidentifyidenticaltwinsonthebasisoftheirfacesandDNAsequences.Inthisresearchpaper,wehaveintroducedanewapproachforidentifyingidenticaltwinsonthebasisofminutiaecoordinates,orientationangles,andminutiaedistancesoftheirthumbprintimages.WetestedtheproposedmethodonthumbprintimagesofanidenticaltwinpairgeneratedbyusingInceptH3T&ATerminalandfiftypairsofidenticaltwinsofFVC04,andFVC06datasets.Wehavefoundthattheproposedapproachissuperior,androbustincomparisontoexistingtechniquesintermsofaccuracy,efficiency,andstoragespace.IndexTerms—FingerprintPatterns,FraternalTwins,IdenticalTwins,MinutiaeDistances,MinutiaeOrientationFieldsI.INTRODUCTIONBiometricsystemsareautomatedmethodsofverifyingapersononthebasisofhisphysiologicalandbehavioralcharacteristics.Physiologicalfeaturesbasedbiometricsincludeirisimage,fingerprints,thumbprint,palmprint,face,fingerveins,handgeometry,DNAsequence,andpalates.Behavioralfeaturesbasedbiometricsincludegaits(styleofwalking,andstyleoftalking)signatureandvoice.Physiologicalandbehavioralcharacteristicscanbeusedtoidentifyapersonifthesecharacteristicssatisfyuniversality,distinctiveness,permanence,andcollectabilityconditions[12].Thetwinsareclassifiedintotwocategoriesnamelyfraternaltwins(monozygotictwins)andidenticaltwins(zygotictwins).Fraternaltwinsareproducedfromtwoeggsandtheseeggsarefertilizedseparatelybytwodifferentsperms.Therefore,fraternaltwinsdifferinfacestructure,DNAsequenceandvoice[6].Itiseasytoidentifyfraternaltwinsonthebasisoffacestructure,DNAsequence,andvoice.Identicaltwinsarefertilizedfromthesameegg.Therefore,identicaltwinshavethesameDNAsequenceandfacestructure.Identicaltwinsidentificationisalwaysachallengingtask.Identicaltwinsdifferinthumbprints(TP),handgeometry(HG),irisimages(II),fingerveins(FV),palatalstructure(PS),fingerprints(FP)andgaitsignatures(GS).Therefore,eitheroneofthesebiometricsortheircombinationscanbeusedforidenticaltwinsidentification[14][16].Mathematically,anIdenticalTwinIdentificationSystem(ITIS)canbedefinedasfollow:52MinutiaeDistancesandOrientationFieldsBasedThumbprintIdentificationofIdenticalTwinsCopyright©2013MECSI.J.Image,GraphicsandSignalProcessing,2013,2,51-59(TP,HG,II,FV,PS,FP,GS)∈ITISITIS={TPHGIIFVPSFPGS}(1)IfITISusesacombinationoftwoormorebiometricsforidentificationthenitcanberepresentedas:ITIS=mCnwhere,2≤n≤m(2)Here,mrepresentsthetotalnumberofbiometrics,andnrepresentsthenumberofbiometricsusedforidenticaltwinsidentification.Figure1:Facialimagesofidenticaltwinpair(SatyaPrakash-DevPrakash).Thefacialimagesofanidenticaltwinpairaregiveninfigure1.Theidenticaltwinsoffigure1aresharingsamechromosomesandsimilarphysicalcharacteristics.Therefore,theycannotbedistinguishedusingDNAsequenceorfacialstructure.Athumbprintbasedverificationandidentificationofidenticaltwinswaspresentedby[8].Inverificationphase,thesystemcomparesaninputthumbprintwiththeenrolledthumbprintofaspecificusertodetermineiftheyarefromthesamefinger.Inidentification,thesystemcomparesaninputfingerprintwiththeprintsofallenrolledusersinthedatabasetodetermineifthepersonisalreadyknownunderaduplicateorfalseidentity.Inthisresearchpaperwehaveusedanewtechniqueforidentifyingthumbprintimagesofidenticaltwinsonthebasisofminutiaecoordinateswithorientationanglesandminutiaedistances.II.RELATEDWORKIdenticaltwinshavetheclosestgeneticsbasedrelationship.Therefore,themaximumsimilaritiesbetweenfingerprintsareexpectedinidenticaltwinpairs.AnilK.Jainshowedthatastate-of-the-artautomaticfingerprintidentificationsystemcansuccessfullydistinguishidenticaltwinpairsonthebasisoftheirfingerprintswithaslightlyloweraccuracy[6].AnilK.Jainproposedanewmethodforidentifyingidenticaltwinsusingautomaticfingerprintverificationsystem.Thesystemcansuccessfullydistinguishidenticaltwinsonthebasisofminutiaeorientationfields.Thematchingalgorithmtransformstwominutiaesetstoacommonframeofreferencebyusingtranslationandrotationbasedalignmentprocedure.Finally,thematchingscoresarecalculatedasafunctionofthenumberof„matched