An user identity authentication scheme adaptive to changes in face appearance

Digital identities (i.e. DIDs) are often used to access e-mails, e-banks, e-shops, physical environments, shared computers, and other restrict access environments. Nowadays, the simple access control methods like passwords or PINs can not be considered sufficiently secure, because an impostor can obtain and use these information without the user knowing it. To make these systems more reliable, multi-modal DID authentication techniques combining different verification steps have been proposed. Biometric features are one of the most reliable alternatives for the authentication problem because, theoretically, they are unique for each person. In this paper, we propose to perform the authentication task by integrating the traditional login information (e.g. username and password) with an adaptive face analysis scheme. A low complexity facial feature extraction method is presented along with our proposed face appearance management method, which could work with more complex facial representation schemes. The preliminary experimental results are encouraging, since 98% of the identities are correctly verified using our proposed authentication scheme, while different face appearances of the same subjects are learned with a relatively small computational effort (i.e. with or without glasses, beard, hat, different poses, etc.).