A multilevel structural technique for fingerprint representation and matching

Incorporation of global features in minutia-based fingerprint recognition schemes enhances their recognition capability but at the expense of a substantially increased complexity. In this paper, we introduce a novel low-complexity multilevel structural technique for fingerprint recognition by first decomposing a fingerprint image into regions based on only some of the global features and then formulating multilevel feature vectors to represent the fingerprint by employing both the global and local features. A fast multilevel matching algorithm based on the new fingerprint representation is proposed. In order to show the effectiveness of the proposed scheme, extensive experiments are conducted using challenging benchmark databases from the 2002, 2004 and 2006 Fingerprint Verification Competitions (FVC2002, FVC2004 and FVC2006), and the results compared with those of some state-of-the-art schemes. The experimental results show that the average template size of the fingerprint representation is only 253 bytes, whereas the average enrollment and matching time is about 0.23 s. The proposed scheme is shown to yield recognition accuracy higher than that provided by the existing schemes at a lower cost.

► A low-complexity multi-level structural scheme for fingerprint recognition is proposed.
► A new fingerprint representation using both the global and local features is presented.
► A fast multi-level matching algorithm based on the new representation is developed.
► Experiments are conducted using benchmark databases from FVCs 2002, 2004 and 2006.
► The proposed scheme provides higher recognition accuracy with lower cost.