Face Anti-spoofing using Eyes Movement and CNN-based Liveness Detection

Biometric authentication is more and more popular these days. Among authentication techniques, face recognition is the most widely used technique. Face anti-spoofing is the core of the biometric system. Face Anti-spoofing is about prevention of spoofing attack by detecting the face image is live or not before feeding it to the system. In this paper, we propose two streamed line approaches for face anti-spoofing. First approach is detecting of eyes movement and second approach is CNN-based liveness detection by extracting the local features. Experiment demonstrates on comparison of previous works with HTER (Half Total Error Rate) over three datasets: NUAA imposter dataset, Replay Attack and OWN replay dataset which is created in this paper.