Fingerprint matching with cross correlation and minutiae scores

This paper reports a combination technique for fingerprint identification (matching). The proposed method utilizes both Normalized cross correlation (NCC) template matching and minutiae matching. This method offers better performance when compared to each individual matching. Moreover, our method is less sensitive to rotational effect. The proposed technique is, however, based on the availability of core point. The technique fails if the core point cannot be located or incorrectly located. The first matching is performed based on the NCC score matching of the pattern around core point. Only few candidates are again matched based on the observed minutiae (i.e., bifurcate and end points with a particular range of predefined angle). Using this 2-step matching, we can improve the FAR and FRR. Images under tested are those obtained in FVC-2004 (DB-1A). There are 16 templates (4 groups of 4 patterns each). We have used 1,000 specimens (100 patterns for 10 times) in our experiment. Upon the investigation, we have obtained 0% FAR and 1% FRR.