指纹识别算法的研究与实现-翻译.doc

指纹识别算法的研究与实现-翻译.doc

《指纹识别算法的研究与实现-翻译.doc》由会员分享，可在线阅读，更多相关《指纹识别算法的研究与实现-翻译.doc（24页珍藏版）》请在冰点文库上搜索。

指纹识别算法的研究与实现-翻译

苏州大学本科生毕业设计（论文）附件:

外文文献资料与中文翻译稿

外文文献资料

收集:

苏州大学应用技术学院10电子,学号1016405029,,李磊,

FINGERPRINTRECOGNITIONUSINGMINUTIASCOREMATCHINGABSTRACT:

ThepopularBiometricusedtoauthenticateapersonisFingerprintwhichisuniqueandpermanentthroughoutaperson’slife.Aminutiamatchingiswidely

usedforfingerprintrecognitionandcanbeclassifiedasridgeendingandridgebifurcation.InthispaperweprojectedFingerprintRecognitionusingMinutiaScoreMatchingmethod（FRMSM）.ForFingerprintthinning,theBlockFilterisused,whichscanstheimageattheboundarytopreservesthequalityoftheimageandextracttheminutiaefromthethinnedimage.Thefalsematchingratioisbettercomparedtotheexistingalgorithm.

Key-words:

-FingerprintRecognition,Binarization,BlockFilterMethod,MatchingscoreandMinutia.

1.Introduction

Biometricsystemsoperateonbehavioralandphysiologicalbiometricdatatoidentifyaperson.Thebehavioralbiometricparametersaresignature,gait,speechandkeystroke,theseparameterschangewithageandenvironment.Howeverphysiologicalcharacteristicssuchasface,fingerprint,palmprintandirisremainsunchangedthroughoutthelifetimeofaperson.Thebiometricsystemoperatesasverificationmodeoridentificationmodedependingontherequirementofanapplication.Theverificationmodevalidatesaperson’sidentitybycomparingcapturedbiometricdatawithreadymadetemplate.Theidentificationmoderecognizesaperson’sidentitybyperformingmatchesagainstmultiplefingerprintbiometrictemplates.Fingerprintsarewidelyusedindailylifeformorethan100yearsduetoitsfeasibility,distinctiveness,permanence,accuracy,reliability,andacceptability.Fingerprintisapatternofridges,furrowsand

minutiae,whichareextractedusinginkedimpressiononapaperorsensors.Agood

第1页

苏州大学本科生毕业设计（论文）附件:

外文文献资料与中文翻译稿

qualityfingerprintcontains25to80minutiaedependingonsensorresolutionandfingerplacementonthesensor.Thefalseminutiaearethefalseridgebreaksduetoinsufficientamountofinkandcross-connectionsduetooverinking.Itisdifficulttoextractreliablyminutiafrompoorqualityfingerprintimpressionsarisingfromverydryfingersandfingersmutilatedbyscars,scratchesduetoaccidents,injuries.MinutiabasedfingerprintrecognitionconsistsofThinning,Minutiaeextraction,MinutiaematchingandComputingmatchingscore.Motivation:

Themotivationbehindtheworkisgrowingneedtoidentifyapersonforsecurity.Thefingerprintisoneofthepopularbiometricmethodsusedtoauthenticatehumanbeing.TheproposedfingerprintverificationFRMSMprovidesreliableandbetterperformancethantheexistingtechnique.

Contribution:

InthispaperweusedFingerprintRecognitionusingMinutiaScoreMatchingmethodwiththehelpofMATLABcodes.Minutiaeareextractedfromthethinnedimageforbothtemplateandinput

image.Finallyboththeimagesaresubjectedtomatchingprocessandmatchingscoreiscomputed.

Organization:

Thispaperisorganizedintothefollowingsections.SectionIIisandefinitionoftherelatedworkanddescribesModelforfingerprintrecognitionindetail,SectionIIIgivesthealgorithm.InsectionIVperformanceanalysisandresultsarediscussedandfinallyinsectionVgivetheconclusions

2.Relatedwork

G.SambasivaRaoetal.,proposedfingerprintidentificationtechniqueusingagraylevelwatershedmethodtofindouttheridgespresentonafingerprintimagebydirectlyscannedfingerprintsorinked36impression.RobertHastingsdevelopedamethodforenhancingtheridgepatternbyusingaprocessoforienteddiffusionbyadaptationofanisotropicdiffusiontosmooththeimageinthedirectionparalleltotheridgeflow.Theimageintensityvariessmoothlyasonetraversealongtheridgesorvalleysbyremovingmostofthesmallirregularitiesandbreaksbutwiththeidentityoftheindividualridgesandvalleyspreserved.JinweiGu,

etal.,proposedamethodforfingerprintverificationwhichincludesbothminutiaeandmodelbasedorientationfieldisused.Itgivesrobustdiscriminatory

第2页

苏州大学本科生毕业设计（论文）附件:

外文文献资料与中文翻译稿

informationotherthanminutiaepoints.Fingerprintmatchingisdonebycombiningthedecisionsofthematchersbasedontheorientationfieldandminutiae.V.VijayaKumariandN.SuriyanarayananproposedamethodforperformancemeasureoflocaloperatorsinfingerprintbydetectingtheedgesoffingerprintimagesusingfivelocaloperatorsnamelySobel,Roberts,Prewitt,CannyandLoG.Theedgedetectedimageisfurthersegmentedtoextractindividualsegmentsfromtheimage.RajuSonavane,andB.S.Sawantpresentedamethodbyintroducingaspecialdomainfingerprintenhancementmethodwhichdecomposesthefingerprintimageintoasetoffilteredimagesthenorientationfieldisestimated.Aqualitymaskdistinguishestherecoverableandunrecoverablecorruptedregionsintheinputimagearegenerated.Usingtheestimatedorientationfield,theinputfingerprintimageisadaptivelyenhancedintherecoverableregions.EricP.Kukula,etal.,purposedamethodtoinvestigatetheeffectoffivedifferentforcelevelsonfingerprintmatchingperformance,imagequalityscores,andminutiaecountbetweenopticalandcapacitancefingerprintsensors.Threeimageswerecollectedfromtherightindexfingersof75participantsforeachsensingtechnology.Descriptivestatistics,analysisofvariance,andKruskal-Wallisnonparametrictestswereconductedtoassesssignificantdifferencesinminutiaecountsandimagequalityscoresbasedontheforcelevel.Theresultsrevealasignificantdifferenceinimagequalityscorebasedontheforcelevelandeachsensortechnology,yetthereisnosignificantdifferenceinminutiaecountbasedontheforcelevelsofthecapacitancesensor.Theimagequalityscore,showntobeeffectedbyforceandsensortype,isoneofmanyfactorsthatinfluencethesystemmatchingperformance,yettheremovaloflowqualityimagesdoesnotimprovethesystemperformanceateachforcelevel.M.R.Girgisaetal.,proposedamethodtodescribeafingerprintmatchingbasedonlinesextractionandgraphmatchingprinciplesbyadoptingahybridschemewhichconsistsofageneticalgorithmphaseandalocalsearchphase.Experimentalresultsdemonstratetherobustnessofalgorithm.LupingJi,andZhangYiproposedamethodforestimatingfourdirectionorientationfieldbyconsideringfoursteps,i）preprocessingfingerprintimage,ii）determiningtheprimaryridgeoffingerprintblockusingneuronpulsecoupledneuralnetwork,iii）estimatingblockdirectionbyprojectivedistancevarianceofaridge,insteadofafullblock,iv）correcting

第3页

苏州大学本科生毕业设计（论文）附件:

外文文献资料与中文翻译稿

theestimatedorientationfield.DuoqianMaioetal.,usedprincipalgraphalgorithmbykegltoobtainprincipalcurvesforautofingerprintidentificationsystem.Fromprincipalcurves,minutiaeextractionalgorithmisusedtoextracttheminutiaeofthefingerprint.Theexperimentalresultsshowscurvesobtainedfromgraphalgorithmaresmootherthanthethinningalgorithm.AlessandraLumini,andLorisNannidevelopedamethodforminutiaebasedfingerprintanditsapproachtotheproblemastwo-classpatternrecognition.TheobtainedfeaturevectorbyminutiaematchingisclassifiedintogenuineorimposterbySupportVectorMachineresultingremarkableperformanceimprovementXifengTongetal.,proposedamethodtoovercomenonlineardistortionusingLocalRelativeErrorDescriptor（LRLED）.Thealgorithmconsistsofthreestepsi）apairwisealignmentmethodtoachievefingerprintalignmentii）amatchedminutiaepairsetisobtainedwithathresholdtoreducenon-matchesfinallyiii）theLRLED–basedsimilarity

measure.LRLEDisgoodatdistinguishingbetweencorrespondingandnoncorrespondingminutiae-pairsandworkswellforfingerprintminutiaematching.

Lametal.,presentedamethod,thinningistheprocessofreducingthicknessL.

ofeachlineofpatternstojustasinglepixelwidth.Therequirementsofagoodalgorithmwithrespecttoafingerprintarei）thethinnedfingerprintimageobtainedshouldbeofsinglepixelwidthwithnodiscontinuitiesii）Eachridgeshouldbethinnedtoitscentralpixeliii）Noiseandsingularpixelsshouldbeeliminatediv）nofurtherremovalofpixelsshouldbepossibleaftercompletionofthinningprocess.Mohamedetal.,presentedfingerprintclassificationsystemusingFuzzyNeuralNetwork.Thefingerprintfeaturessuchassingularpoints,positionsanddirectionofcoreanddeltaobtainedfromabinarisedfingerprintimage.Themethodisproducinggoodclassificationresults.Ching-TangHsiehandChia-Shing–Hu[14]hasdevelopedanoidmethodforFingerprintrecognition.Ridgebifurcationsareusedasminutiaeandridgebifurcationalgorithmwithexcludingthenoise–likepointsareproposed.Experimentalresultsshowthehumanoidfingerprintrecognitionisrobust,reliableandrapid.

LieWeiproposedamethodforrapidsingularitiessearchingalgorithmwhichusesdeltafieldPoincareindexandarapidclassificationalgorithmtoclassifythefingerprintinto5classes.Thedetectionalgorithmsearchesthedirectionfieldwhichhasthelargerdirectionchangestogetthesingularities.Singularities

第4页

苏州大学本科生毕业设计（论文）附件:

外文文献资料与中文翻译稿

detectionisusedtoincreasetheaccuracy.HartwigFronthaler,etal.,[Proposedfingerprintenhancementtoimprovethematchingperformanceandcomputationalefficiencybyusinganimagescalepyramidanddirectionalfilteringinthespatialdomain.ManaTarjomanandShaghayeghZareiintroducedstructuralapproachtofingerprintclassificationsbyusingthedirectionalimageoffingerprintinsteadofRavi.J.et