Using 3D Models for Improving Face Recognition

Abstract

Face recognition algorithm Principal Component Analysis (PCA) has a significant performance drop when comparing photographs taken from different angle. In this paper a 3D model was used for improving that performance. Model enables us to transform the face image which is taken from certain angle to en face. Model has been tested against biometric database formed at the Faculty of Organizational Sciences. Image rotation based on the model was performed before matching with the en face images from the database. Study results show that algorithm precision on biometric verification and identification has been seriously improved.