FPGA-based embedded hand vein biometric authentication system

Abstract

Biometric authentication provides a high security and reliable approach to be used in security access system. However, this authentication method has not been widely implemented in a resource-constrained embedded system. In this project, we investigate a method of personal authentication based on infrared vein pattern in the back of the hand, targeted for embedded system which is implemented in Altera Nios II FPGA prototyping system, running on Real Time Operating System (RTOS). The RTOS applied is Nios2-Linux. A biometric feature is extracted from the vein pattern image and then matched for personal authentication. The algorithm consists of four modules: image capturing, image pre-processing, feature extraction, and the authentication module. These image processing algorithms executed in an embedded system with a 50 MHz fixed point processor. Preliminary experiment on a database of 82 images of 15 candidates show promising result with the system reaching equal error rate of 5.5%.