Implementation of iris recognition technique for E-security

• Main Author: Loke, Jonathan Xiang
• Other Authors: Mohammed Yakoob Siyal
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/69299
• Institution: Nanyang Technological University

Iris recognition is touted to be one of the most accurate, reliable and promising form of biometrics. Developed by Dr. John Daugman, Daugman’s iris recognition algorithms are used in most commercial iris recognition systems today. Due to the decreasing cost of iris recognition technology, there are more commercial research, development of new algorithms and applications of iris recognition than before. In addition, as password authentication alone is increasingly deemed insufficient, there is a push towards multi-factor authentication in the field of e-security, As a result of the two forces converging, there exists an opportunity and desire to integrate biometric authentication, so as to improve the reputation of e-commerce security. With a more secure e-commerce landscape, online merchants can benefit from more customers choosing e-commerce over regular commerce. Since iris recognition is not the only form of biometric authentication, the other biometrics are also studied and compared in this project to weigh their respective advantages and disadvantages. By implementing Daugman’s algorithm using MATLAB software and publicly available database of iris images, the author modelled an iris recognition system. Experiments were then conducted to simulate the usage of an actual iris recognition system for e-security purposes, as closely as possible. Results from the implemented MATLAB program were then analyzed to evaluate whether it worked according to theory and if the iris recognition error rates are as low as reported. Finally, given the shortcomings faced, works were recommended to improve future implementations of iris recognition systems. Even though iris recognition technology itself has several theoretical advantages, thanks to the uniqueness of irises, malicious attackers will inevitably find ways to circumvent iris recognition. Hence, whenever a human factor is added to the equation, even a promising technology like iris recognition cannot provide a miraculous, 100% fool-proof authentication method for e-security. One can only strive to improve iris recognition, algorithm-wise (technically) and implementation-wise (managerially), to achieve reasonably practicable e-security.