Multimodal Biometric Template Transformation Approach using a List Ranking Algorithm

Abstract

User authentication has become vital with the advancement in technologies that comes with an increased threat to security. Biometrics is typically used to secure the data of individuals based on their unique features, such as fingerprints, retinas, and hand geometry. Template transformation is a technique in which the original feature vectors are transformed into modified feature vectors. When these templates are transformed and stored, it becomes difficult for imposters to hack the system. Existing systems store and secure these templates based on key generation. When the key information is compromised, the system can be easily hacked. Hence, in the proposed system, the templates are transformed using a modified list ranking algorithm and then stored for verification. Also, in order to improve security, instead of using a single biometric, the proposed system obtains biometric data from left fingerprints, right fingerprints and palm prints and fuses them into a single feature vector set. Thus, even when the repository is attacked, it would be very difficult to break into the system. The proposed system provides a 2.9 % equal error rate compared with existing systems.