Face spoofing detection and Authentication Using Linear Binary Patterns, Gabor Features and Support Vector Machine

Abstract

Face detection and authentication have become an active field of research material in recent years duo to the increased use of face dependent access control systems, which can be considered as a good alternative to other biometric features such as fingerprint duo to its easily accessible, non- intrusive nature that makes it very important during the ongoing pandemic years. However, this method doesn’t come without security risks related to adversaries seeking to gain unauthorized access to the system. Face spoofing is a security breach attempt occurs when the attacker tries to deceive the face-enabled access control system by displaying a photo, video or wearing a mask of an authorized person to gain access. The paper in hand suggests a method for face anti-spoofing by utilizing some of the well-known feature extraction techniques usually associated with face detection and recognition, namely the LBP and Gabor features, in addition to the commonly used SVM classifier for identifying real and spoofed feces. The algorithm is implemented successfully on multiple individuals, achieving performance levels comparable to other accredited methods in terms of FAR, FRR and HTER verification criteria, aspiring for more advancedand effective methods.