FinCaT: a novel approach for fingerprint template protection using quadrant mapping via non-invertible transformation.

The employment of various technologies for secured human authentication in the recent years has led to rapid expansion of biometric-based recognition. Among all biometrics, the fingerprint recognition systems (FRS) hold the largest market share and have been used in numerous computing applications such as forensic, governance, and securing critical infrastructure. Though biometric systems provide plentiful benefits over the traditional identification systems but these are also susceptible to various adversarial attacks. The most crucial among all the attacks involves security breaches of stored templates in a central database that may either degrade the overall performance or result in a complete failure of the FRS. To countermeasure these attacks, template protection mechanisms are designed that mitigate the masquerade attacks or template thefts. In this study, we present a novel cancellable approach for fingerprint template protection (FinCaT) using the novel notion of quadrant mapping via a non-invertible transformation function. Our method transforms original fingerprint templates to highly secured templates by using quadrant mapping that maps minutia points by using distinct parameters in each quadrant. The FinCAT approach yields high revocability and diversity as the compromised template of a genuine user can be replaced with a newly generated secured template. The approach is experimentally evaluated on the FVC 2002 fingerprint benchmark dataset. Our technique demonstrates decent performance in terms of an equal error rate (EER) of 5.95% and a recognition accuracy of 94.05%, which is promising as compared to similar state-of-the art template protection techniques. [ABSTRACT FROM AUTHOR]