Secure Access To Authorized Resources Based On Fingerprint Authentication

• Main Author: Elmadani, Ahmed Baba
• Format: Thesis
• Language: English; English
• Published: 2003
• Online Access: http://psasir.upm.edu.my/id/eprint/12158/1/FK_2003_17_A.pdf; http://psasir.upm.edu.my/id/eprint/12158/

The Internet makes it convenient for anyone to access publicly available information on the servers. The increased functionality of computers and other technological equipment makes the connectivity easier than ever before, and the need for security more and more important. Passwords are frequently used to control access to restricted functions. Unfortunately, password or personal identification number (PIN) verification suffers an inherent problem. It cannot ensure that the user is the claimed individual. Higher security systems have now veered towards biometric verification in conjunction with passwords. Fingerprints are a practical bodily characteristic to use, as they are unique to each individual and easily collected using image-capture systems. In this thesis, the methods of fingerprint recognition and classification are investigated. Then, the possible approaches to use are discussed while investigating the subject. The final choice combines a feature-based and correlation-based approach. The thesis proposes a novel method of allowing users access after authenticating them by their fingerprints. The method is based on a statistical approach, and is crafted in such a way that the authentication operations are inconspicuous to the user. All the required image processing techniques that make the extraction of the true fingerprint features easier are used: equalization, filtering, binarization and thinning. A new method for constructing a unique key from the fingerprint image is also presented, with the fingerprint database (fingerprint features, unique fingerprint key, public and private keys) created shown. The database can be manipulated (insertion, retrieving, and deletion) rapidly using the Adelson Velskii and Landis (AVL) tree searching technique. The A VL tree is used to increase the compression ratio as its compression algorithm works efficiently for all types of data. The security is maintained and enhanced by adding a public key system. Finally, a unique fingerprint based key is constructed using a user's fingerprint image and the key used to access and retrieve the private key. The method successfully recognizes the user fingerprint image even with noisy, translated or rotated images.