SIFT特徴量を用いた顔認証に関する一考察

Abstract

There is ID card as the method of personal authentication. This is not necessarily a surefire way in terms of safety, forgery and loss. This is why, biometric authentication using physical features is gathering attention. Face authentication is a part of it. But there are some problems when face image are hidden by masks and sunglasses. In such case, face would not be necessarily succeeded with conventional authentication methods. SIFT(Scale Invariant Feature Transform) is one of the algorithms that describes features of brightness, orientation and size invariant features. We think SIFT features would be able to solve the problem in face authentication as above conditions. Goal of our study is personally identifiable using face image in that some parts, such as eye or nose, are invisible. The purpose of this paper is to examine the possibility of using SIFT features in face authentication.