Development of Core-based Minutiae Triplets Model for an Improved Fingerprint Matching

Identifying distorted fingerprint images is a major problem in fingerprint recognition system. Distorted images can be caused by a change in scale of fingerprint images at capturing stage and at identification or verification stage. In this paper, distortion that may arise due to change in scale is addressed using an improved matching algorithm described as corebased minutiae triplets. This proposed algorithm uses minutiae triplets as a basis for fingerprint matching. For each minutiae point, a minutiae triplet is created which includes the core point and the closest minutiae to the core point. To make the proposed algorithm more effective, some optimizations were included so as to discard non-matching minutia triplets without comparing the whole representation. From experimentations carried out using the same data set, results showed that the core-based minutiae triplets algorithm had an average percentage accuracy of 98.25% while the minutiae inter-distance-based algorithm by had an average percentage accuracy of 55.5%. Therefore, the proposed algorithm in the paper outperformed the minutiae inter-distance-based algorithm in terms of accuracy by 42.75%.