Fake Face Distinction System by Using Monocular Camera and Method thereof

摘要

A forged face discriminating method using a monocular camera of the present invention comprises the steps of: storing 3D reference model information of a person or an object; receiving a face image from a camera; detecting a face area from the received face image; determining whether the detected area is a face; if the detected area is a face, extracting feature points of the face; determining whether a current face (t) is identical to a previous face (t-1) on the basis of the extracted feature points; if the current face is identical to the previous face, estimating a 3D face model for the current face; checking whether estimating the 3D face model for the current face is processed by N frame; on the basis of eyes coordinates among the feature points of the estimated 3D face model processed by N frame, generating an accumulated 3D estimated model through aligning and merging; and comparing similarity between the accumulated 3D estimated model and pre-stored 3D reference model information to determine whether the face is forged and outputting the determination result. The forged face discriminating method can be applied to an existing face authentication system without additional sensors to discriminate a forged face in real time. Accordingly, the cost for building a system can be reduced by using a monocular camera and high reliability can be secured by reducing the complexity of the system.