Deep Learning-Based Biometric Finger Vein Authentication System for Enhanced Security

Abstract

Biometrics, which uses human physiological characteristics, is a method for protecting personal information. Recently, finger vein authentication has become one of the most popular biometric techniques. This method offers high security and accuracy, making it a reliable form of biometric authentication. The system compares a person's vascular structure in their finger to previously collected data. Finger vein authentication works by identifying vein patterns beneath the skin's surface. The proposed system aims to enhance user authentication security by leveraging the uniqueness of finger vein patterns. The finger vein image is obtained from a database, and preprocessing is done using a Gaussian median filter in both spatial and frequency domains to remove noise. Image segmentation is performed through a line tracking method, which enhances image contrast. For feature extraction, the system utilizes Convolutional Neural Networks (CNN), and these features are matched with the stored finger vein database. A deep learning approach is then applied to classify users as genuine or imposters. In real-time, a scanner captures the finger vein image, which is sent to an Arduino board for storage and subsequently processed in MATLAB for classification. The result is transmitted through a GSM module as an alert or message, and the information is also stored in an IoT system for future reference. A GSM module is integrated with the user for communication. The proposed system achieves an accuracy of 96%, making it highly beneficial for security applications like access control, identity verification, banking, and financial transactions.