Robust Presentation Attack Detection in Finger Vein Recognition using ECS-LTP And LPQ Methods

This research focuses on the development of a Presentation Attack Detection method for Finger Vein Recognition systems. Finger vein biometrics is an emerging technology that utilizes the unique vascular patterns within the finger for personal recognition. However, finger vein systems remain vulnerable to impersonate attacks by using forged images, commonly refer to as presentation attacks. This study incorporates two robust texture-based PAD methods, namely Extended Centre Symmetric Local Ternary Pattern and Local Phase Quantization. The proposed methods are evaluated using SCUT-FVD datasets from South China University of Technology and VERA from Idiap Research Institute, achieving lowest Average Classification Error Rate of 0.12% and 0%, respectively