A Novel Tree Based Shuffling Technique for Fingerprint Template Security

摘要

Person Identification is the important issue in the modern computing environment. Biometric recognition systems provide enormous user support for authentication services instead of the tradition authentication systems such as token based or password methods and so on. But still a hacker can steal the template information and reconstruct the original biometric trait of the authenticated user. In order to solve this issue, many encryption approaches such as key binding approaches and key generation approaches are proposed. All these methods have their own merits and demerits with respect to providing template security. Less error rate with improved recognition and security are the key challenging issues in most of the existing models. This paper proposes a novel transformation approach based on a complex user defined tree, which overcomes most of the drawbacks of the existing systems. This method extracts the feature vectors and perform some kind of shuffling operation based on dynamic user generated tree logic and store the completely reordered feature sets into the database. From the hacker perspective it is very hard to reconstruct the original fingerprints from the template database. Moreover, the proposed method achieves 1.37 % Equal Error Rate (EER). Hence, the proposed system improves the security without affecting the traditional matching performance.