A Review on Vein Biometric Recognition: Techniques, Datasets, and Models

Abstract

Vein biometric recognition has emerged as a promising and, above all, reliable security solution, thanks to its resistance to tampering and the physiological stability of vein patterns, even over time. Among the various modalities, hand and finger veins offer a rich source of information and relevant data for reliable individual identification and authentication. To enhance the performance of this recognition modality, a lot of image processing, machine learning techniques and methods have been developed.

This paper provides a comprehensive review of the main approaches and recent techniques used for vein feature extraction, focusing on traditional methods such as Gabor filters and local binary patterns (LBP), as well as more recent and revolutionary techniques based on convolutional neural networks (CNNs). Some of the most commonly used databases in this field are also presented, as well as the various recognition models and methods that exploit these features. This study aims to better understand current advances, performance levels, and remaining challenges in vein-based biometric recognition.