Study of Contactless Finger-Vein Identification System

A deep learning-based finger-vein identification system is introduced to detect fingers’ rotational and translational movements as well as interference from external lights. In addition, the effects of multiple preprocessing methods and the system’s effectiveness are investigated. The results demonstrate that the proposed system’s higher identification accuracy is 98.1 % for the SDUMLA-HMT public database. Finally, the system’s high accuracy and stability, and contactless applications indicate its practicality in midst of the COVID-19 pandemic.