Presentation attack detection on visible spectrum iris recognition by exploring inherent characteristics of Light Field Camera

Abstract

Presentation (or spoof) attacks on biometric system is a growing concern that received substantial attention from both academics and industry. In this paper, we present a novel way of addressing a Presentation Attack Detection (PAD) (or spoof detection) by exploiting the inherent characteristics of the Light Field Camera (LFC) for visible spectrum iris biometric system. The proposed PAD algorithm will capture the variation in the depth (or focus) between multiple depth images rendered by the LFC that in turn can be used to reveal the presentation attacks. To this extent, we introduce a new presentation attack database comprised of 52 subjects with 104 unique eye samples. The database is collected using LFC by simulating the attacks through visible spectrum iris biometric artefacts like printed photo and electronic display (using both Apple iPad (4 th generation) and Samsung Galaxy Note 10.1 tablet). Extensive experiments carried out on this database reveal the efficacy of the proposed PAD algorithm with a lowest Average Classification Error Rate = 0.5% when confronted with diverse set of attacks on visible spectrum iris biometric system.

Introduction to Presentation Attack Detection in Iris Biometrics and Recent Advances

“…Photo attacks are probably the presentation attacks against IRS most studied in the literature [13,48,50,54,58,63]. They simply consist on presenting to the sensor an image of the attacked iris, either printed on a sheet of paper or displayed on a digital screen.…”

“…Nonetheless, it is relatively easy to capture a video in public spaces and the usage of long range sensors open allows video capturing at a distance with high resolution qualities. • Expected impact of the attack: The literature offers a large number of approaches with good detection rates of printed photo attacks [13,48,50,54,58,72]. However, most of these methods exploit the lack of realism/quality of printed images in comparison with bona fide samples.…”

“…hemoglobin presents an absorption peak in Near-Infrared Bands). • 3D imaging [43,54]: The curvature and 3D nature of the eye has been exploited by researchers to develop PAD methods. The 3D profile of the iris is captured in [43] by using two Near-Infrared light sources and a simple 2D sensor.…”

“…The idea underlying the method is to detect the shadows on real irises produced by non uniform illumination provided from two different directions. Light Field Cameras (LFC) are used in [54] to acquire multiple depth images and detect the lack of volumetric profiles of photo attacks. • Electrooculography [42]: The electric standing potential between the cornea and retina can be measured and the resulting signal is known as electrooculogram.…”

Presentation Attack Detection for Face Recognition Using Light Field Camera

“…With this backdrop, in our recent work [21] we presented a preliminary study on presentation attack detection on visible spectrum iris recognition. Our preliminary results carried out on adopting LFC for PAD on visible iris recognition motivated us to extend this work in many directions.…”