Automated Real-time Door Access Control using Finger-vein Recognition

Abstract

The article focuses on developing an automated and real-time finger-vein identification device for door access control. This device records an individual's finger-vein image to grant access to authorized users. The door opens if the user is authorized and remain closed otherwise. The finger-vein image is captured by a Raspberry Pi camera (CMOS NOIR) when a matrix of near-infrared light-emitting diodes (NIR LEDs) illuminates the finger inside the acquisition box. In the captured image, the finger or the background appears lighter, while the veins appear darker. The image then undergoes several processing stages to enhance its quality before the Modified Hausdorff Distance (MHD) technique compares the minutiae points with the template stored in the database. The 2D Entropy algorithm performs further analysis to identify the correct user. If the user is authorized, the door opens; otherwise, it remains closed. The device is developed using Raspberry Pi 3 as the microcontroller, which processes the image, controls the camera, manages NIR LED lighting, and operates the door motor. The device achieves an Equal Error Rate (EER) of 7.88%, corresponding to an accuracy of 92.12%. This study's contribution includes detailed development specifics and proposed solutions to issues encountered during the research.