Extraction of Stable Points from Fingerprint Images Using Zone Could-be-in Theorem

摘要

This paper presents a novel zone Could-be-in theorem, and applies it to interpret and extract singular points (cores and deltas) and estimate directions of cores in a fingerprint image. Where singular points are regarded as stable points (attracting points and rejecting points just according to their clockwise or anticlockwise rotation), and pattern zones are stable zones. Experimental results validate the theorem. The corresponding algorithm is compared with popular Poincare index algorithm under two new indices: reliability index (RI) and accuracy cost (AC) in FVC2004 datasets. The proposed algorithm are higher 36.49% in average RI, less 2.47 in average AC, and the advantage is more remarkable with the decrease of block size.