基于手指静脉和手指轮廓的个性化融合方法.pdfVIP

计算机研 究与发展 ISSN 1000—1239／CN 111777／TP JournalofComputerResearchandDevelopment 50(9)：1914—1923，2013 基于手指静脉和手指轮廓的个性化融合方法 袭肖明 尹义龙 杨公平 孟宪静 (山东大学计算科学与技术学院 济南 250101) (fyzql0@126．corn) PersonalizedFusionM ethodBasedonFingerVein andFingerContour XiXiaoming，YinYilong，YangGongping，andMengXianjing (SchoolofComputerScienceandTechnology，ShandongUniversity，Jinan250101) Abstract Fingervein isapromising biometricfortheauthentication due to itssomeadvantages． However，whencapturingthefingerveinimage，theposevariation ofthefingerortheillumination variationmaycausefailuretoverification．Toovercometheselimitationsofusingasinglebiometric， multiplebiometricscanbecombinedtoimprovetheperformance．Comparedwiththefusionofother biometrics，theadvantageofthefusion offingervein and fingercontouristhatacquiring the tWO biometricimageisconvenientbecausethefingerveinimageandfingercontourimagecan beobtained by afingervein capturing device．In thispaper，weproposeapersonalized fusionmethodbasedon fingerveinandfingercontouratthescoreleve1．Thesamplesarefirstlyclassifiedbasedontheoriginal matching score，and then the classification confidencescore (CCS) can beobtained based on the classification result． Compared with the traditional score，CCS contains additional classification informationwhich may providemoreusefulinformation forthefinalfusion．In addition，amore effectivepersonalized weight fusion based on CCS isproposed due to the differenceofdifferent subjects．Finally，weconductextensiveexperimentsonourdatabasetoevaluatetheeffectivenessof ourproposa1． Keywords fingervein recognition；fingercontourrecognition；multimodalbiometricsverification； classificationconfidencescore(CCS)；personalizedweightfusion 摘 要 近年来手指静脉识别成为一种有前途的身份验证技术．采集图像 时，手指的姿 势或光照的变化 会严重影响算法的性能．多生物特征识别可以在一定程度上克服这些限制 ，提高识别性能．相 比其他的 生物特征融合，手指静脉和手指轮廓融合的优点是采集图像比较方便 ，只需要手指静脉采集设备 即可获 得这两种生物特征的图像