Abstract: Biometric recognition refers to recognizing a person based on one or more of his/her physical (fingerprint, face, iris, hand geometry etc.) or behavioral (signature, voice, keystroke dynamics etc.) traits. Biometric traits are intrinsically associated with persons and cannot be forgotten or shared with others. Physical traits are stable, unique and do not change significantly in time as compared to behavioral traits. However, one of the most significant vulnerabilities of biometrics is that once a biometric template is compromised, it cannot be canceled and reissued. An attacker could then gain access to all the applications using that same biometric trait. In this paper, a new approach is proposed for protection of iris template using behavioral trait. Behavioral trait is only used to generate user specific key. It presents an effective combination of biometrics data with user specific biometric key for human recognition to generate cancelable and non-invertible biometric templates. Neither the key nor the image can be recovered from the resulting template. The gain obtained from the proposed approach is two-fold: cancelability and non-invertibility. Experimental evaluations are performed on a public dataset demonstrating the accuracy of proposed approach.
Keywords: Iris biometric, voice biometric, biometric template, security, user specific key, cancelability.