Robustness of multimodal biometric fusion methods against spoof attacks

摘要: In this paper, we address the security of multimodal biometric systems when one modes is successfully spoofed. We propose two novel fusion schemes that can increase systems. The first an extension likelihood ratio based scheme and other uses fuzzy logic. Besides matching score sample quality score, our proposed also take into account intrinsic each system being fused. Experimental results have shown methods are more robust against spoof attacks compared with traditional methods.