An Efficient Hand-Based Biometric Recognition System Using Finger- Knuckle-Print Data
Abdallah Meraoumia, Salim Chitroub and Ahmed Bouridane
Affiliation: Universite Kasdi Merbah Ouargla, Laboratoire de Genie Electrique., Faculte des Sciences et de la Technologie et des Sciences de la Matiere, Ouargla, 30000, Algerie.
Keywords: Biometrics, Identification, Finger-Knuckle-Print (FKP), 2D Block based Discrete Cosine Transform (2D-BDCT),
Multivariate Normal density (MVN), Sum of Absolute Differences (SAD), Log-likelihood, Data fusion.
Automatic personal identification is playing an important role in secure and reliable applications, such as access
control, surveillance systems, information systems, physical buildings and many more applications. In contrast with
traditional approaches, based on what a person knows (password) or what a person has (tokens), biometric based identification
is providing an improved security for their users. Biometrics is the measurement of physiological traits such as
palmprints, fingerprints, iris etc., and/or behavioral traits such as gait, signature etc., of an individual person for personal
recognition. Hand-based person identification provides a good user acceptance, distinctiveness, universality, relatively
easy to capture and low-cost. However, Finger-Knuckle-Print (FKP), which provides different information from a variety
of finger types, has been recently used to improve the performance of hand-based biometric identification because each
finger has a specific feature, making it possible to collect more information to improve the accuracy of hand-based biometric
systems. In this paper, we presented an efficient online personal identification based on FKP using the twodimensional
Block based Discrete Cosine Transform (2D-BDCT) and MultiVariate Normal density function (MVN). In
this study, a segmented FKP is firstly divided into non-overlapping and equal-sized blocks, and then, applies the 2DBDCT
over each block. By using zig-zag scan order, each transform block is reordered to produce the feature vector. Subsequently,
we use the MVN for modeling the feature vector of each FKP. Finally, Log-likelihood scores are used for FKP
matching. Finally, performance of all finger types is determined individually and several fusion rules are applied to develop
a multimodal system based on fusion at the matching score level. Experimental results show that FKPs modalities
show best performance for identifying a person as they provide an excellent identification rate and with more security.
Here we also discuss few patents that are relevant to the article.
Rights & PermissionsPrintExport