Fingerprint matching algorithm

Abstract

This project report explains and investigates a non-core based fingerprint minutiae matching algorithm and the classification of the fingerprints into different categories. Fingerprint matching is the decision making part of a fingerprint identification system and thus it must be reliable and efficient enough to be used in a real-time application. A non-core based fingerprint minutiae matching algorithm is used in this project because it eliminates the strenuous process of aligning the fingerprints to be matched, does not run the risk of selecting a false core point and carries the potential to cater for almost all the inconsistencies in the representation of the fingerprint data. The algorithm devised in this project takes care of most of these inconsistencies: translation, rotation, missing points, additional points and noise. The algorithm is reliable and the computational complexity of the algorithm is O(n2k) that makes it ideal to be used in real-time fingerprint matching applications. Matching a fingerprint takes time. Therefore, the less the number of fingerprints there are to be compared for a match the faster the application would be. Classification is used to divide the fingerprint database into subsets of categories such that when matching, only a subset of the fingerprint database is used for comparison thus making the algorithm faster. There are three levels of classification devised in this report: Physical, Finger type and Fingerprint level. The implementation of the algorithm investigated in this report only uses the Fingerprint level of classification and shows promising results of the practicality of the algorithm.