Fingerprint orientation field extraction using gradient-based weighted averaging

Abstract

This paper presents a new algorithm for extracting the fingerprint orientation field (FOF). The basic idea of the algorithm is to compute the ridge orientation by combining weighted averaging with angle doubling for fingerprint orientation field extraction. The similarities of point orientations are first computed, and then the weight values of point gradients can be obtained according to the similarity. In the second procedure, to calculate the local ridge orientation, the angles of gradient vectors are doubled and the weighted averaging of moduli of gradient vectors is implement. The FOF is extracted by iterating weighted averaging based on overlapped window. The experiments with the database of FVC2004 are carried out using various FOF extraction methods. The experimental results show, compared to other extraction algorithms, the accuracy and reliability of the FOF extracted by our method have improved to some extent.