Robust multimodal face and fingerprint fusion in the presence of spoofing attacks

Peter Wild,Petru Radu,Lulu Chen,James Ferryman

doi:10.1016/j.patcog.2015.08.007

Abstract

Anti-spoofing is attracting growing interest in biometrics, considering the variety of fake materials and new means to attack biometric recognition systems. New unseen materials continuously challenge state-of-the-art spoofing detectors, suggesting for additional systematic approaches to target anti-spoofing. By incorporating liveness scores into the biometric fusion process, recognition accuracy can be enhanced, but traditional sum-rule based fusion algorithms are known to be highly sensitive to single spoofed instances. This paper investigates 1-median filtering as a spoofing-resistant generalised alternative to the sum-rule targeting the problem of partial multibiometric spoofing where m out of n biometric sources to be combined are attacked. Augmenting previous work, this paper investigates the dynamic detection and rejection of liveness-recognition pair outliers for spoofed samples in true multi-modal configuration with its inherent challenge of normalisation. As a further contribution, bootstrap aggregating (bagging) classifiers for fingerprint spoof-detection algorithm is presented. Experiments on the latest face video databases (Idiap Replay-Attack Database and CASIA Face Anti-Spoofing Database) and fingerprint spoofing database (Fingerprint Liveness Detection Competition 2013) illustrate the efficiency of proposed techniques.

Highlights

Fingerprint and face biometrics as most widely adopted traits are being exposed to an increasing threat of presentation attacks
This paper investigates 1-median filtering as a spoofing-resistant generalised alternative to the sum-rule targeting the problem of partial multibiometric spoofing where m out of n biometric sources to be combined are attacked
As a second outlined contribution, this paper presents a novel multibiometric spoofing-aware fusion method following the idea of anomaly detection and extending research in [10] to multiple modalities

Summary

Introduction

Fingerprint and face biometrics as most widely adopted traits are being exposed to an increasing threat of presentation attacks. Recog nition scores can be helpful in the probe-attack spoofing detection problem and liveness scores can impact on the recognition task. Considering imposters with access to fake fingers or face photographs reveals an impact on overall accuracy (shifted imposter score distribution for non-zero-effort attempts [7]) and assuming a correlation between successful spoofs achieving a higher score and their 15 corresponding liveness score is likely (and shown) to help in the final judgment of the decision task, especially in an ensemble of classifiers where this paper looks for outliers. The paper focuses on three objectives: (1) investigation of spoofing robustness in 30 multibiometrics; (2) development of novel methods towards anomaly detection for increased systematic anti-spoofing; and (3) proposition of a novel bootstrap aggregating (bagging) of classifiers method combining features in fingerprint counter-spoofing

Methods

Findings

Discussion

Conclusion