A variant of Biohashing based on the chaotic behavior of the Logistic Map

To meet the security threats related to biometric authentication such as fraudulent access, biometric forgery and identity theft, a number of protection systems have in fact emerged. One of the most famous and widely used in the context of cancellable biometrics is the so-called Biohashing. This two-factor authentication technique has shown great promise in terms of recognition accuracy and biometric template security. However, its critical flaw is that it becomes unpractical when a genuine token is stolen, since performance in this situation is getting extremely poor. To address this issue, we propose in this paper a robust variant of Biohashing technique which is dedicated to facial authentication security. The main concept of this proposal consists in making Biohashing depend on a sequence of random numbers generated by the Logistic Map instead of the one presented by the token. The experiments tested on the Yale face database have shown satisfactory results in both performance and security even under the stolen-token scenario.