A few techniques for fingerprint template protection

Abstract

The role of authentication systems has turned out to be exceptionally vital in the present automated world. Conventional authentication systems rely on ID cards and secret information. While these procedures are extremely prominent, they have a few constraints and hindrances, for example, the watchword of a client can effortlessly be overlooked, stolen or speculated and the ID cards can be lost, stolen, shared or harmed. These shortcomings can be overcome by the use of authentication systems based on biometrics. Biometrics deals with recognizing a user based on his/her physiological or behavioral characteristics. By applying sensors, both physiological (such as fingerprint, iris, face, ear etc.) as well as behavioral (such as signature, handwriting, voice, etc.) characteristics can be captured from the user. These characteristics are relatively unique, permanent and difficult to forge and share. Also, a user does not need to worry about forgetting them or making any special effort to carry them along. Among various biometric traits (features), fingerprint is the most widely used trait for user authentication. The use of fingerprint is becoming popular day by day and many countries are currently using it as a prime tool for the authentication of their citizens.Fingerprints provide several advantages; however, there are many issues related to privacy and security in fingerprint-based authentication systems. The authentication systems relying on fingerprint usually use minutiae points information and store it directly as a user template in the database. It has been proven from various studies that from the information of minutiae points, the original fingerprint can be reconstructed. Databases are prone to get attacked by an adversary, and in the case of fingerprint-based authentication system, if the fingerprint features (minutiae points) of a human being are compromised (through an attack by the adversary), then they cannot be replaced or revoked. As fingerprint can be generated using the information of its minutiae points, it is essential to ensure the security of the fingerprint data in biometric systems. Primarily, the work presented in the thesis addresses the security issues related to fingerprint based user templates. It proposes various efficient techniques for the computation of a secure fingerprint based user template. The user template generated using the proposed techniques show good resistance against various attacks and no information about the original fingerprint features can be obtained (revealed) from the user template generated by these techniques. The user template computed by these techniques fulfills the essential requirements such as revocability, diversity, security, and performance of a secure and robust template. Outcomes of the experimental analysis conducted on FVC2002 DB1, FVC2002 DB2, and FVC2002 DB3 fingerprint databases show highly encouraging performance and exhibit the viability of the proposed techniques. The presented techniques have also been analyzed with respect to various attacks and are found to be highly secure and robust.