الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 142 |  |  |
| | | from | 142 | | |
Abstract
Despite their usability advantages over traditional authentication systems, biometrics-based authentication systems suffer from inherent privacy violation and non-revocability issues. Due to the intra-user variations that exist in biometric data, several fuzzy encryption schemes have been proposed in the last few years to protect biometric templates and preserve users’ privacy. These fuzzy encryption systems can be broadly classified into two main categories; namely cancelable biometric (CB) and biometric cryptosystems (BCs). Unfortunately, existing CB schemes require integrating other authentication factors such as password keys and/or user specific tokens with biometric data. Such multi-factor authentication techniques suffer from the same issues associated with traditional knowledge-based and token-based authentication systems. On the other hand, BCs, in which a user-specific cryptographic key is linked to or extracted from his/her biometric data such that the key is released only if a genuine sample is presented at the time of verification, are not designed to be revocable with respect to biometric templates. Moreover, the length of the key to be linked with or extracted from a biometric template is constrained by the size of that template. This dissertation tackles the above issues and provides innovative solutions. First, a reliable tokenless CB scheme, called BioEncoding, for securing iris codes, is proposed. Although BioEncoding does not require user-specific keys or passwords to be employed in the cancelable transformation process, it satisfies the CB requirements of revocability, diversity and noninvertibility without deteriorating the recognition performance of the original unprotected biometric system. Moreover, the transformation process of BioEncoding is easy to implement and can be integrated simply with current iris matching systems. The effectiveness of the proposed method is confirmed experimentally using CASIA-IrisV3-Interval dataset. Second, the security of BioEncoding, in terms of both non-invertibility and privacy protection, is analyzed thoroughly in this thesis. First, resistance of protected templates generated using BioEncoding against brute-force search attacks is discussed rigorously. Then, vulnerabilities of BioEncoding with respect to correlation attacks and optimization based attacks are identified and explained. Several approaches are proposed to enhance the security of BioEncoding against different attacks. Finally, a highly secure hybrid fuzzy encryption scheme that integrates CB and BCs effectively is presented. The proposed hybrid scheme addresses the main issues associated with current BCs. First, using the proposed system, cryptographic keys of unlimited length can be secured using the derived protected templates rather than the limited size iris codes. Additionally, revocability of cryptographic keys as well as true iris templates is guaranteed due to the hybridization of both techniques. Experimental results show that the proposed hybrid system can achieve perfect recognition accuracy (0% ERR) regardless of the key size.