Anti-spoofing using challenge-response user interaction

Abstract

2D facial identification has attracted a great amount of attention over the past years, due to its several advantages including practicality and simple requirements. However, without its capability to recognize a real user from an impersonator, face identification system becomes ineffective and vulnerable to spoof attacks. With the great evolution of smart portable devices, more advanced sorts of attacks have been developed, especially the replayed videos spoofing attempts that are becoming more difficult to recognize. Consequently, several studies have investigated the types of vulnerabilities a face biometric system might encounter and proposed various successful anti-spoofing algorithms. Unlike spoofing detection for passive or motionless authentication methods that were profoundly studied, anti-spoofing systems applied on interactive user verification methods were broadly examined as a potential robust spoofing prevention approach. This study aims first at comparing the performance of the existing spoofing detection techniques on passive and interactive authentication methods using a more balanced collected dataset and second proposes a fusion scheme that combines both texture analysis with interaction in order to enhance the accuracy of spoofing detection.