Flexible Bimodal Recognition of Collaborators in Pervasive Environments

We propose an access control method based on the combination of two biometric recognitions: voice and face. In particular, our face recognition algorithm aims at determining a person's identity when he/she is involved in situations in which his/her face is rotated in a shoulder to shoulder trajectory, which is a common behavior on people with a stealth or intromission attitude. The other biometrics, the voice recognizer, allows us to confirm the prediction made by the face recognizer. Our method can be integrated into several systems, as it has been developed as a Web application and it does not require any special hardware, so it is highly flexible and multi-platform at the same time. Experiments reveal that our proposed method performs really well, as it gives 89% of accuracy, unlike traditional face recognition algorithms, which could not identify anyone, whose head presents a pronounced inclination.