الفهرس | Only 14 pages are availabe for public view |
Abstract ABSTRACT Fingerprint recognition is one of the commonly-used biometric recognition schemes. With the advances in hacking technologies and tools, it has become possible for intruders to break the security of institutions and capture vital information from their databases. Biometric databases have become a target for these types of attacks. Due to the confidential nature of biometric databases as being composed of unique signatures for subscribers, the attacks on these databases are very serious. It is not possible to use a biometric again if it is stolen. It is also not possible to change the individual biometric. So, fingerprint encryption has captured the attention of researchers in the last decade. Encrypted fingerprints can take place of the original fingerprints in the so-called cancellable fingerprint recognition. The thesis begins with fingerprint pre-processing with enhancement tools to improve the quality of latent fingerprints. After that, four fingerprint encryption schemes are presented and tested with encryption quality metrics. These schemes are tested also in the cancelable fingerprint recognition scenario. They prove high verification rates. Finally , one of the well-known concepts in communication theory, which is aliasing concept, is utilized for fingerprint security. This is achieved by sampling of fingerprints beyond the Nyquest rate, and hence interpolation of these fingerprints to create an intended distortion of fingerprint templates that is not reversible or removable. These distorted fingerprint templates are used in the cancellable fingerprint recognition scenario. High verification rates are guaranteed with this new aliasing-based scenario. Finally, a multi-biometric cancellable recognition scheme based on DCT compression and aliasing is presented. It guarantees compactness of storage of multiple biometric and security in verification through the aliasing concept. This scheme succeeded in obtaining high recognition rates. |