3D Mask Face Anti-spoofing with Remote Photoplethysmography
3D mask spoofing attack has been one of the main challenges in face recognition. Among existing methods, texture-based approaches show powerful abilities and achieve encouraging results on 3D mask face anti-spoofing. However, these approaches may not be r
- PDF / 1,281,776 Bytes
- 16 Pages / 439.37 x 666.142 pts Page_size
- 94 Downloads / 146 Views
Department of Computer Science, Hong Kong Baptist University, Kowloon Tong, Hong Kong {siqiliu,pcyuen}@comp.hkbu.edu.hk 2 School of Computer Science and Technology, Harbin Institute of Technology, Harbin, China [email protected] 3 Center for Machine Vision and Signal Analysis, University of Oulu, Oulu, Finland [email protected]
Abstract. 3D mask spoofing attack has been one of the main challenges in face recognition. Among existing methods, texture-based approaches show powerful abilities and achieve encouraging results on 3D mask face anti-spoofing. However, these approaches may not be robust enough in application scenarios and could fail to detect imposters with hyper-real masks. In this paper, we propose a novel approach to 3D mask face antispoofing from a new perspective, by analysing heartbeat signal through remote Photoplethysmography (rPPG). We develop a novel local rPPG correlation model to extract discriminative local heartbeat signal patterns so that an imposter can better be detected regardless of the material and quality of the mask. To further exploit the characteristic of rPPG distribution on real faces, we learn a confidence map through heartbeat signal strength to weight local rPPG correlation pattern for classification. Experiments on both public and self-collected datasets validate that the proposed method achieves promising results under intra and cross dataset scenario.
Keywords: Face anti-spoofing plethysmography
1
·
3D mask attack
·
Remote photo-
Introduction
Face recognition has been widely employed in a variety of applications. Like any other biometric modality [1,2], a critical concern in face recognition is to detect spoofing attack. In the past decade, photos and videos are two popular media of carrying out spoofing attacks and varieties of face anti-spoofing algorithms have been proposed [1–12] and encouraging results have been obtained. Recently, with the rapid development of 3D reconstruction and material technologies, 3D mask attack becomes a new challenge to face recognition since affordable off-the-shelf c Springer International Publishing AG 2016 B. Leibe et al. (Eds.): ECCV 2016, Part VII, LNCS 9911, pp. 85–100, 2016. DOI: 10.1007/978-3-319-46478-7 6
86
S. Liu et al.
(a) Genuine Face
(b) Masked Face
Fig. 1. Effect of remote photoplethysmography (rPPG) on normal unmasked face (a), and masked face (b). (a) shows rPPG on a genuine face: Sufficient light penetrate the semi-transparent skin tissue and interact with blood vessels. rPPG signal can go through skin and be detected by RGB camera. (b) depicts rPPG on a masked face: The mask material blocks large portion of the light that the skin should absorb. Light source needs to penetrate a layer of painted plastic and a layer of skin before interacting with the blood. Remain rPPG signals will be too weak to be detected
masks1 have been shown to be able to spoof existing face recognition system [13]. Unlike the success in traditional photo or video based face anti-spoofing, very few methods have been proposed to address 3D mask face
Data Loading...
