Multibiometric Cryptosystems Based on Feature-Level Fusion

摘要: Multibiometric systems are being increasingly de- ployed in many large-scale biometric applications (e.g., FBI-IAFIS, UIDAI system India) because they have several advantages such as lower error rates and larger population coverage compared to unibiometric systems. However, multibiometric require storage of multiple templates fingerprint, iris, face) for each user, which results increased risk user privacy security. One method protect individual is store only the secure sketch generated from corresponding template using a cryptosystem. This requires sketches. In this paper, we propose feature-level fusion framework simultaneously single sketch. Our main contributions include: (1) practical implementation proposed two well-known cryptosystems, namery,fuzzy vault fuzzy commitment, (2) detailed analysis trade-off between matching accuracy security cryptosystems based on different databases (one real one virtual multimodal database), containing three most popular modalities, namely, face. Experimental show that both here higher performance their counterparts.