Paper Title
Multimodal Biometric Authentication System Using A Feature-Level Fusion of Dorsal Hand, Palm and Finger Veins

As technology and innovation have advanced,thedemandformorereliableauthenticationhasgrowninrecent years. In particular, the use of biometric authenticationhasbecomeincreasinglypopularduetoitsaccuracyandconvenience. The Traditional biometric authentication systemsdepend on a single physical characteristic, like a fingerprint orface recognition.However, thesemethods aresusceptible tospoofingand othertypesof attacks.Tosolvethisproblem,multimodalbiometricauthenticationthatcombinesseveralfeatures has been proposed as asolution.In thisstudy, weintroduceamultimodalbiometricauthenticationtechniquethat fuses dorsal, finger and palm vein images at feature levelusing local binary pattern (LBP), principal component analysis(PCA) and k-nearest neighbor (KNN) classifiers. The proposedsystem extracts the dorsal hand, palm and finger veins imagesfrom a single scan of the user’s hand. Images are preprocessedtoeliminatethenoiseandenhancefeaturesbeforebeingextractedusingthelocalbinarypatterntechnique. Theprincipal component analysis is applied to combine elementsfrom local binary patterns to decrease the size of the featurevector. Lastly, authentication is achieved by implementing a K-Nearest Neighbor (KNN) classifier. The proposed method isevaluatedusingtheSDUMLAT-HMTdatabase,theVERAPalm vein database, and the Pontificia Universidad Javeriana'scollection of Dorsal Hand Vein Images. The proposed systemoutperforms all other existing methods with 98% accuracy anda minimum EER of 0.01%. The proposed method provides aneffective approach to user authentication. It is very reliable,accurate and effective. Additionally, it can precisely distinguishbetweenlegitimate users andimposterusers. Keywords:Biometricauthentication,MultimodalBiometricSystem, dorsal and vein,palm vein,fingerveins,principalcomponentanalysis(PCA),featureextraction,localbinarypatterns(LBP),fusion,classifier,Knearestneighbor( KNN).