Abstract
Biometric technologies have gained much interest recently. The identities of individuals are recognized by directly utilizing their physiological or behavioral characteristics. This removes the problems of conventional techniques such as forgetting passwords or losing keys. However, despite many advantages offered by these technologies, concerns about compromising individuals' privacy are accompanied for the reasons as follow. First, most of the biometric traits (e.g., fingerprints and faces) in use are extrinsic, which may be easily recorded without the owners’ consent. Moreover, the biometrics of an individual is permanently associated with him/her so that it is difficult to be revoked when stolen or compromised. Several attempts have been made to address these concerns, and "cancelable biometrics" is the one attracted the most attention. To reflect this trend, an ECG based cancelable biometric scheme is proposed in this study. Experiments with real ECGs showed that our proposed scheme achieved a recognition rate of 97.19%. Furthermore, the biometric templates generated by our proposed scheme fulfill all the requirements to be cancelable, including revocability, non-invertibility, and diversity.