Facial Landmark Features-Based Face Misclassification Detection System

Issues of face spoofing that can evade the verification system by placing the photo of real user on camera have been discussed a lot in the literature survey. By detecting the person through misclassification, the problem could be minimized. Therefore, in this paper, robust face misclassification detection system is proposed using ABT mechanism. The proposed system provides the additional level of security before face recognition module. Face landmark features such as eye, nose, and mouth movements are used for generating challenges for detecting fake users from genuine users using misclassification. The reliability of system is tested by placing photographs and videos from Replay-Attack database and live database. Proposed system gives good results under spoofing attacks such as eye imposter attack and mouth imposter attack. The results show that system detects the fake user when implemented on all types of attacks and confirms the 79.6% misclassification detection.