Towards Secure Personal Device Unlock using Stereo Camera Pan Shots

Personal mobile devices hold sensitive data and can be used to access services with associated cost. For security reasons, most mobile platforms hence implement automatic device locking after a period of inactivity. Unlocking them using approaches like PIN, password or an unlock pattern is both problematic in terms of usability and potentially insecure, as it is prone to the shoulder surfing attack: an attacker watching the display during user authentication. Therefore, face unlock - using biometric face information for authentication - was developed as a more secure as well as more usable personal device unlock. Unfortunately, when using frontal face information only, authentication can still be circumvented by a photo attack: presenting a photo/video of the authorized person to the camera.
We propose a variant of face unlock which is harder to circumvent by using all face information that is available during a 180° pan shot around the user's head. Based on stereo vision, 2D and range images of the user's head are recorded and classified along with sensor data of the device movement. We evaluate different classifiers for both grayscale 2D and range images and present our current results based on a new stereo vision face database.