Fingerprint retrieval approach based on novel minutiae triplet features

Fingerprint retrieval technique is necessary to reduce the search space in case of applications with huge size fingerprint database. A novel fingerprint retrieval approach based on minutiae triplets is proposed to speed up the fingerprint matching in this paper. The number of matched minutiae polygons can be derived from matching information of minutiae triplets to offer additional information to evaluate the similarity of two minutiae sets, and a one-bit flag for matching score weighting is attached to each index to enhance the matching precision. Experimental results on FVC 2000, 2002, 2004 databases show that the proposed approach outperforms the state-of-the-art systems. Experimental results on NIST SD27 with a large background database demonstrate the good performance of the proposed approach even on difficult latent-to-roll filtering. Our proposed technique significantly increases the accuracy and efficiency of fingerprint retrieval in large databases.