A novel approach for curvature detection in global fingerprint feature extraction

Fingerprint features extraction plays a critical role on any Automatic Fingerprint Identification System (AFIS). Extracting curvatures from the fingerprint accurately is a critical step in AFIS and optimizing its computation without losing the resolution could be very beneficial as it is for any process in applications consisting of image processing. In curvature extraction progressing in variable steps shows advantages versus the current fixed step methods. Experimental results show the proposed approach could improve the truthfulness of the extracted curves as well as the process time. As anticipated, the difference is more visible in high curvature ridges.