Robust Fingerprint Verification Using M-Triplets

Abstract

Fingerprint verification has become one of the most active research areas nowadays. A key component of an accurate fingerprint verification system is the fingerprint matching algorithm. An accurate matching algorithm uses a robust fingerprint representation. In this paper, we introduce m-triplets, a new minutiae triplet representation and similarity for fingerprint verification. The proposed similarity shifts the triplets to find the best minutiae correspondence and it uses rules that discard not matching minutiae triplets without comparing the whole representation; hence, achieving high matching speed. To test the quality of the introduced representation and similarity, we modify a popular fingerprint verification algorithm. The modified algorithm achieves the best accuracy and speed in all the databases of FVC2004 compared with four accurate verification algorithms.