An Innovative Biometric Model: Deep Representation through Feature Extraction for Finger Vein Extraction

Abstract

Considerable evidence suggests that developments in vein-vein pattern identification and verification have improved the effectiveness of biometric validation. However, existing solutions are still vulnerable to manipulation at the raw data source. The study presents an innovative finger-vein biometric model consisting of deep representation-based feature extraction for enhanced reliability.  Noticeable improvements occur in the identification and verification efficacy by training Convolutional Neural Network (CNN) for understanding the differences between vein patterns and their background. Further, the study developed the first of its kind Fully Convolutional Network (FCN). The experimental result shows that the proposed method produces strong vein patterns, significantly improving fingerprint verification accuracy, which will be useful for various organizational functions.