Fusion of finger types for fingerprint indexing using minutiae quadruplets

• Fingerprint indexing performance is enhanced by fusion.
• Fingerprint types are fused at the rank level using the highest rank rule.
• Fingerprints are indexed using minutiae quadruplet features.
• The clustering retrieval strategy is employed.
• Minutiae quadruplets are viable features and show good performance from the experiments.

Accuracy, security and efficiency are the drivers for exploiting multibiometric systems. A multi-modal or multibiometric system is relatively more secure than a single-biometric system as a result of the rich biometric content, lower error rates and resilience to impostor attacks. The availability and easy acquisition of the ten prints of an individual with fewer resources, give multi-finger biometric fusion scheme an advantage over other fusion schemes based on different biometric modalities. In this paper, multi-finger indexing is proposed using minutiae quadruplets in combination with a clustering scheme. Four, five and ten fingerprints from a subject are fused at the rank level using the highest rank rule. The minutiae quadruplet features are observed to be robust and the clustering scheme assists in quickly identifying a list of potential candidates in the gallery database. Results from the experiments show good performance of the fusion scheme with minutiae quadruplet features in fingerprint indexing.