An improved authentication system using hybrid of biometrics and cryptography

• Published in: 2017 IEEE 3rd International Conference on Electro-Technology for National Development (NIGERCON) pp. 457 - 463
• Main Authors: Oluwadamilola, K. Oyetola, Ayodeji, A. Okubanjo, Martins, O. Osifeko, Olufunmi, I. Sanusi, Rapheal, O. Abolade
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2017
• Subjects: Authentication; Biometric; Biometric templates; Cryptography; Data Security; Face; Feature extraction; Fingerprint recognition; Iris recognition
• ISSN: 2377-2697
• DOI: 10.1109/NIGERCON.2017.8281915

Data security is indispensable in our society, as human activities are increasingly becoming computerized and are transferred through complex communication networks across the globe. Existing methods used in securing data include cryptography and biometrics. However, the main shortcoming of cryptography is poorly-chosen or forgotten password while challenges with biometrics include interclass similarities in the feature sets used to represent trait and storing of biometric templates in clear unprotected format. This paper presents combination of cryptography and biometrics; a bimodal biometric Cryptosystem, using fingerprint and face as trait for authentication. Subjects' information were encrypted using Advanced Encryption Standard (AES) and biometric templates were stored as Binary Large Object (BLOB) in MYSQL database secured with Message Digest 5 (MD 5) Hashing Algorithm. The system was developed and implemented to operate on one-try, two-try and three-try configurations at varying threshold values. The developed system's performance was evaluated using False Reject Rate (FRR), False Accept Rate (FAR) and Receiver Operating Characteristic Curve (ROC graph) as performance metrics. On ROC graph, three-try configuration gave optimal performance at all threshold values.