An embedded Gabor-based palm vein recognition system

Biometric authentication systems are a growing field of research that represent an alternative to the more common authentication systems and often offer better performances. Among the available technologies the palm vein recognition represents a non invasive, reliable and affordable solution. Thanks to its appreciated peculiarities, this technology is carving a spot in the market of biometric recognition both at the consumer and business level. In this paper we describe the development of an embedded standalone palm vein authentication system. In order to realize a contact-less acquisition system, a solid ROI-locating algorithm is proposed. A combination of Gabor filters and histograms calculations is investigated as a method for creating biometric templates. The prototype has been developed using the low-cost single board computer Raspberry Pi 3 and it is thought to handle all the steps, from the acquisition of the image of the hand, to the recognition of the user. The entire authentication process is completed in 1.5-2s, with a FAR of 0.32%, a FRR of 1.58% and an EER of 1.45%. Moreover, thanks to its low cost and to the on site extraction of the biometric template, the proposed device perfectly fits as terminal of a bigger authentication system, with multiple access points and a central system where the matching and storage occur.