Recently, a walkthrough type vein authentication system has been attracting extensive attention, which is effective for wide-scale events such as big event venue, theme park, and so on. In our previous study, a hand waving finger vein authentication system was proposed, in which a similarity between enroll and verification finger vein patterns was calculated based on Normalized Cross Correlation [1] and Scale-Invariant Feature Transform [2]. These methods perform worse in case that the baekground of the captured frame has significant noises such as the bright light sources. In order to eliminate such noises and extract a finger region correctly, we apply a machine learning method. In this study, we employ U-Net [3] as a machine learning method and evaluate it compared to conventional image processing techniques.

© 2019 Japan Society of Applied Physics, The Optical Society (OSA)

PDF Article


You do not have subscription access to this journal. Citation lists with outbound citation links are available to subscribers only. You may subscribe either as an OSA member, or as an authorized user of your institution.

Contact your librarian or system administrator
Login to access OSA Member Subscription