Sensor-Assisted Facial Recognition: An Enhanced Biometric Authentication System for Smartphones

Facial recognition is a popular biometric authentication technique, but it is rarely used in practice for device unlock or website / app login in smartphones, although most of them are equipped with a front-facing camera. Security issues (e.g. 2D media attack and virtual camera attack) and ease of use are two important factors that impede the prevalence of facial authentication in mobile devices. In this paper, we propose a new sensor-assisted facial authentication method to overcome these limitations. Our system uses motion and light sensors to defend against 2D media attacks and virtual camera attacks without the penalty of authentication speed. We conduct experiments to validate our method. Results show 95-97% detection rate and 2-3% false alarm rate over 450 trials in real-settings, indicating high security obtained by the scheme ten times faster than existing 3D facial authentications (3 seconds compared to 30 seconds).