Face liveness detection using 3D structure recovered from a single camera

摘要: Face recognition, which is security-critical, has been widely deployed in our daily life. However, traditional face recognition technologies practice can be spoofed easily, for example, by using a simple printed photo. In this paper, we propose novel liveness detection approach to counter spoofing attacks recovering sparse 3D facial structure. Given video or several images captured from more than two viewpoints, detect landmarks and select key frames. Then, the structure recoveredfrom selected Finally, an Support Vector Machine (SVM) classifier trained distinguish genuine fake faces. Compared with previous works, proposed method following advantages. First, it gives perfect results, meets security requirement of biometric systems. Second, independent on cameras systems, works well different devices. Experiments faces versus planar photo warped demonstrate superiority over state-of-the-art methods.