Anti-Spoofing Face-Recognition Technique for eKYC Application

Abstract

This paper presents a module of eKYC (Electronic Know Your Customer) system using face recognition for identification and authentication of an individual from a variety of digital sources. The proposed method implements the Local Binary Pattern Histogram (LBPH) algorithm to solve the face recognition problem. The recognition rate varies under lighting conditions, facial expression, attitude deflection and transformations. The anti- spoofing system is based on a proposed Convolution Neural Network (CNN) based architecture. The accuracy of the proposed system is 95%.