[18p-E215-1~9] 【CS.1】 Code-sharing Session 3.3 & 4.4

▲ [18p-E215-9] Improvement on finger region extraction for hand-waving finger vein authentication

〇Hiroyuki Suzuki¹, Muhammad Arkaan Izhraqi¹, Jumpei Nagata¹, Takashi Obi¹, Takashi Komuro² (1.Tokyo Inst. of Tech., 2.Saitama Univ.)

Keywords:machine learning, biometrics

In our previous study, we proposed a hand waving finger vein authentication system, in which finger region extraction from captured images is effective for verifying the finger vein patterns with high accuracy. In this study, we apply a machine leaning method to extract finger region. With experiments, we employed U-Net as a machine leaning architecture and the results show that the generated function contributes to a good performance of the finger region extraction in terms of accuracy as well as computational time.