A Fingerprint Matching Scheme Based on Gradient Difference Compatibility

Abstract

This paper proposes a new minutiae-basedfingerprint matching algorithm using a variation of the Generalized Hough Transform called MGHT, which allows variations in translation, rotation, scale, and some distortions of the fingerprints, with very low complexity. A simple local structure of the fingerprint is constructed to collect template datafrom a number of minutiae. The matching process is performed by voting for transformation parameters by using target fingerprint local structure to look up similar data in the template. The result with a strong peak in the parameter space is considered as a match. The algorithm is efficient, precise, and robust to noise. The test results on several images demonstrate the effectiveness of this fingerprint matching method.