Minutiae matching using local pattern features

摘要

This paper concerns algorithms related to analysis of fingerprint images in area of minutiae matching. Proposed solutions make use of information about minutiae detected from a fingerprint as well as information about main first order singularities. The use of first order singularities as a reference point makes algorithm of minutiae matching more efficient and faster in execution. Proposed algorithms concern efficient detection of main singularity in a fingerprint as well as optimization of minutiae matching in polar coordinates using main singularity as a reference point. Minutiae matching algorithm is based on string matching using Levenstein distance. Detection of first order singularities is optimized using Poincare's index and analysis of directional image of a fingerprint. Proposed solutions showed to be efficient and fast in practical use. Implemented algorithms were tested on previously prepared fingerprint datasets.