The Finger Vein Recognition Using Deep Learning Technique

Abstract

Finger vein biometrics have gained a lot of attention in recent years because they offer the perfect balance of security and economic viability, with advantages such as being the least susceptible to identity theft because veins are present beneath the skin, being unaffected by ageing of the person, etc. All of these factors make it necessary to create functioning models to meet the ever-increasing need for security. The use of facial recognition and AI-based biometrics, particularly in law enforcement, has become a hot topic because of its inadvertent demographic bias. Biometric bias, on the other hand, has far-reaching consequences that extend into daily use cases. When an ATM transaction or an online banking transaction is compromised by a false positive or negative verification, fraudulent activity is made easier. The study in this research work focused on the difficulty of determining the veins of a fingertip. On two widely used and freely available datasets of finger veins, we applied deep convolutional neural network models to feature extraction. Finger vein recognition has gotten a lot of interest recently as a novel biometric technique. Finger vein recognition might benefit from applying deep learning, an end-to-end approach that has shown promising results in sectors like face recognition and target detection.