Design and Development of Low-Cost Sensor to Capture Ventral and Dorsal Finger Vein for Biometric Authentication

Abstract

Biometrics-based authentication of subjects is widely deployed in several real-life applications. Among various biometric characteristics, the finger-vein characteristic has demonstrated both reliable and highly accurate authentications for access control in secured applications. However, most of these systems are based on commercial sensors, where the image-level data are not available for academic research. In this paper, we present the design and development of a low-cost finger-vein sensor based on a single camera that can capture finger-vein images from dorsal and ventral parts of the finger with high quality. The system consists of multiple near-infrared (NIR) light sources to illuminate the finger from both sides (left and right) and top. The camera in the sensor is also coupled with the custom designed physical structure to facilitate high reflectance of the emitted light and distribute the light uniformly on the finger to capture good-quality dorsal and ventral finger-vein patterns. Extensive experiments are carried out on the data captured using the developed sensor and benchmarked the performance with eight different state-of-the-art (SOTA) algorithms. The results on a large-scale finger-vein dataset demonstrate the need for illumination from both sides (left and right) and from the top of the finger, to capture finger-vein images with high quality that improves the verification performance.