Investigating spoofing attacks for 3d cameras used in face biometrics

Abstract

The unique features of human biometrics made it possible to benefit from these biometrics and use them as authentication methods. On the other hand, fraudulent biometrics are attempting to attack biometrics systems and are threatening the security of these systems. These risks can be avoided using liveness detection techniques, such as: heart rate measurement, pupil tracking, image quality assessment and many other techniques. Most face recognition systems use 2D cameras, where a 3D estimation of the face is derived as an antispoofing method. Not many systems are using 3D cameras for facial recognition, and therefore, its vulnerabilities to spoofing techniques are under-explored. In this paper, our purpose is to assess the 3D camera liveness assurance technique, and propose solutions that strengthens the gaps found in spoofing attack detection. Experiments will be conducted where we will use iFace 300 as a case study and attempt to attack the device using different attacking approaches such as mock face masks and 2D printed face images. We expect to discover weaknesses and strengths in the 3D liveness detection technique, and suggest methods to improve this technique.