Fusion of vein and knuckle print for a finger-based biometric system
Loading...
Date
2019-06
Authors
Ong, Chu Mei
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The finger vein and knuckle print are among the most promising biometric traits that can accurately identify a person because their patterns are unique and distinctive. However, unimodal biometrics have limited usage since no single biometric is sufficiently robust and accurate in real applications. The limitations imposed by unimodal biometrics can be overcome by incorporating multimodal biometrics. Based on the review of previously developed systems, the misalignment of finger during acquisition process, the variation of ROI segmentation and the sensitivity of algorithms towards displacement error are the factors that influence the performance of the system. The objective of this project is to improve the performance and accuracy of the previously proposed methods. So, Canonical Correlation Analysis Network (CCANet) is applied on the bimodal biometric system based on finger vein and knuckle print to address the limitations. Since CCANet considers two-view features of one image, the greyscale images of database are used as the first view features, whereas the second view features are extracted from the original images using Maximum Curvature (MC) and Compound Local Binary Pattern (CLBP), respectively. The matching scores of finger vein and knuckle print, which provided by Normalized Histogram Intersection (NHI) method, are fused using Support Vector Machine (SVM)-based score level fusion method. To find the best parameter setting for the entire proposed system, extensive experiments are conducted on two databases, namely FVFKP_USM and THU-FVFDT2. The experimental results show that the proposed CCANet method has better performance than the other existing methods. The fusion of vein and knuckle print for a finger-based biometric system has an Equal Error Rate (EER) of 0.001077% and a Genuine Acceptance Rate (GAR) of 99.9989%, when the proposed method is tested on THU-FVFDT2 database. In conclusion, the bimodal system based on finger vein and knuckle print has higher GAR and lower EER than finger vein and finger knuckle print unimodal systems.