Publication: Fusion of knuckle and n for vein finger based biometric system
Date
2012-06-01
Authors
Wong, Cheah Min
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Biometric system is a pattern recognition system which involves acquiring biometric image from an individual, feature extraction and matching with the template in database. In order to improve the reliability of the biometric system, most of the unimodal biometric system such as fingerprint recognition system, finger vein recognition system, palm print recognition system and face recognition system are transformed into bimodal biometric system gradually. Fusion in biometric system can take three forms which are single biometric multiple representation, single biometric multiple matchers and multiple biometric system. Fusion of finger knuckle and finger vein biometric system is one of the examples for multiple biometric systems. The finger information can be easily obtained where knuckle and vein existed for each finger. The objective of this project is to develop and design a device and software that can capture both finger vein and finger knuckle simultaneously via a Graphical User Interface. Image acquisition device was designed in a way that it can capture finger vein and finger knuckle simultaneously. Then, a Graphical User Interface was designed and developed to form an interaction between user and the system. The matching technique used was Band-Limited Phase Only Correlation (BLPOC). The respective genuine and imposter score generated by knuckle and vein are fused using Weighted Sum method. Experiments were conducted to optimize the parameters for image matching. The proposed finger vein recognition system has a Genuine Acceptance Rate (GAR) of 84.42% and Equal Error Rate (EER) of 15.58%. The proposed finger knuckle recognition system has a GAR of 92.7% and EER of 7.3%. By using weighted sum method, GAR of 94.71% and EER of 5.29% can be obtained. The EER for the proposed finger vein recognition system is too high for a robust system. However, the proposed knuckle recognition system is robust and reliable. In conclusion, the fusion of knuckle and vein for finger based biometric system has a better GAR and a lower EER as compared to finger vein or finger knuckle recognition system alone.