Enhanced Biometric Security Through Infrared Vein Pattern Recognition

Abstract

Finger vein verification stands out as a highly advanced biometric technology today, offering superior security and convenience by utilizing internal body features. The fingerprint images captured by a web camera under infrared (IR) light not only reveal the vein patterns but also the shadows created by varying thicknesses in finger muscles, bones, and surrounding tissue networks. Finger vein verification has proved to be extremely popular among machine learning and pattern recognition researchers but has not been fully explored for its potential, especially with the availability of high computational power, advancements of various deep learning models, and exponential increase in availability of public data for training these models. This paper details an innovative approach to enhance low-resolution images affected by light and noise from a webcam, followed by segmenting the vein patterns using an adaptive threshold method and comparing them with advanced template matching techniques. Despite suboptimal image quality, our results demonstrate that clear arterial patterns and correct procedures enable effective personal identification. The paper thoroughly reviews finger recognition algorithms, encompassing image acquisition, pre-processing, feature extraction, and pattern matching methods. It also documents novel findings from a critical analysis of comparative strategies, highlighting the proposed system’s effectiveness even in challenging imaging conditions. The project aimed to design and develop a finger vein identification system for a networked environment, showcasing its potential as a leading biometric technology by improving image quality and employing robust matching techniques for reliable personal authentication.