摘要
De-duplicationusingbiometricshasgainedmuchattentionfromresearchcommunitiesasitprovidesauniqueidentityforeachandeveryindividualamongthelargepopulation.De-duplicationistheprocessofremovingtheinstancesofmultipleenrollmentsbythesamepersonusingtheperson’sbiometricdata.Animportantissueinthelarge-scalede-duplicationapplicationsisthespeedofmatchingandtheaccuracyofthematchingbecausethenumberofpersonstobeenrolledrunsintomillions.Thispaperpresentsanefficientmethodtoimprovetheaccuracyoffingerprintde-duplicationinde-centralizedmanner.De-duplicationaccuracydecreasesbecauseofthenoisepresentinthedata,whichwouldcauseimproperslapfingerprintsegmentation.Inthispaper,anattemptismadetoremovethenoisepresentinthedatabyusingbinarizationofslapfingerprintimagesandregionlabelingofdesiredregionswith8-adjacencyneighborhood.Thedistinctfeatureofthistechniqueistoremovethenoisepresentinthedataforanaccurateslapfingerprintsegmentationandimprovethede-duplicationaccuracy.Experimentalresultsdemonstratethatthefingerprintsegmentationrateandde-duplicationaccuracyareimprovedsignificantly.
出版日期
2014年02月12日（中国期刊网平台首次上网日期，不代表论文的发表时间）
