الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 108 |  |  |
| | | from | 108 | | |
Abstract
Biometrics has been widely adopted in various applications that require identification or verification. Biometric system storage is the common attack target for impostors. To address this issue, several biometric protection systems have been proposed to ensure the proper use of biometric templates by protecting them against the misuse by attackers. Biometric Template Protection Schemes are divided into two categorize: Cancelable Biometrics and Biometric Cryptosystem. Most of cancelable biometric schemes depend on storing this parameter in a physical token which case the system to fill into tradition authentication problems such as steal-share- forget this token. The knowledge of this parameter will simplify system attacking like the pre-image attack task in creating an approximation to the original biometric data. Thus, the transformed parameter should be stored in a secure way to prevent pre-image attacker from threatening the security of biometric system. This thesis proposes storing this parameter in an association memory model to prevent these problems. This thesis attempts to apply this solution in cancelable biometrics by introducing two schemes. The first one uses simplified-XOR as a transformed function between original biometric template and its transformed key to create XORed transformed template. Although this scheme is considered a pre-image resistant with a high recognition performance, it should be emphasized that it is still vulnerable to correlation attack to recover the original biometric template by correlating several protected cancelable templates created from the same biometric data. Thus, the second proposed scheme will be suggested to overcome this problem by using binarized convolution process as transformation function instead of XOR operation. The proposed schemes ensure the security and privacy requirements of biometric protection systems. Moreover, results obtained from the experiments conducted on iris biometric databases show promising recognition accuracy.