Edge and Texture Analysis for Face Spoofing Detection
Bahia Yahya-Zoubir, Fedila Meriem, Fatiha Mokdad · 2023
With the increasing use of biometric technology in various applications of daily life, face spoofing detection technology plays a crucial role in securing them against fraudulent access. In this paper, a new face anti-spoofing method is proposed that involves a two-stage feature extraction from the YCbCr and HSV face images. The first stage extracts the Binarised Statistical Image Features (BSIF) from the Canny edge face images, while the second stage extracts the Local Binary Patterns (LBP) features from the input face images. The resulting feature vectors from both stages are concatenated to match their identification before being fed into a Support Vector Machine (SVM) classifier. Experiments on the publicly available MSU Mobile Face Spoof Database (MSU-MFSD) confirm that the proposed method achieves promising results compared to state-of-the-art face anti-spoofing techniques.