Spoof Face Recognition in Video Using KSVM. (2016)

• Record Type: Journal Article
• Authors: Faseela, T.; Jayasree, M.
• Is Part Of: Procedia technology. Volume 24(2016)
• Page Start: 1285
• Page End: 1291
• Publication Date: 2016
• Subjects: Face Recognition -- K-Means -- SVM -- Spoof Detection -- KSVM; Technology -- Congresses; Technology -- Periodicals; Engineering -- Congresses; Engineering -- Periodicals; Engineering; Technology; Conference proceedings; Periodicals; 605
• DOI: 10.1016/j.protcy.2016.05.118
• Languages: English
• ISSNs: 2212-0173

Abstract

Abstract: With the advancement of video and its transmission technologies, video has found its use extensively in surveillance, security, especially in detecting forgery. Unlike still images, video provide large amount of intra-personal variations, making face recognition a significant concept in the field of biometric security. Automatic face recognition is a widely used concept in implementing security, which is also prone to various spoof attacks. Spoof attacks accounts to reproducing a person's face using printed photos or by replaying a video. A large number of face recognition and spoof detection algorithms have been developed for still images, but those concerning videos are less in number. Face recognition from videos and related spoof detections are less explored. This paper deals with countering such spoof attacks in facial recognition using videos, where KSVM (K-Means and SVM) is used to identify the recognized images to be real or spoof. KSVM, a combined concept of K-Means and SVM outperforms simple SVM.