A novel approach on cluster-based indexing technique for level-1 and level-2 fingerprint features

摘要

In recent years, automated fingerprint identification system (AFIS) plays a predominant role in personal authentication and verification. Fingerprint indexing is still a challenging issue in an AFIS as the size of database nowadays is huge. This paper proposes a cluster-based indexing on fingerprint features. The feature extraction of a fingerprint image is at three levels: level-1, level-2 and level-3. Fingerprint indexing is broadly classified as correlation-based matching, minutiae-based matching and non-minutiae-based matching. The fingerprint database is clustered using k-means algorithm. The fingerprint indexing methodology projected in this work is based on a combination of level-1 and level-2 features.