STATISTICAL-METHODS FOR AUTOMATIC INTERPRETATION OF DIGITALLY SCANNEDFINGER PRINTS

Orientation flow field estimation is the starting point for feature ex traction in various fingerprint algorithms. Jain and his colleagues (R atha et al., 1995, 1996; Karu and Jain, 1996) recently introduced a fu ndamentally new algorithm motivated by fingerprint matching. Rather th an identification, our aim has been feature extraction from a dermatog lyphic viewpoint with an interest in other quantitive features, such a s ridge counts and local average ridge widths (see Loesch, 1983), whos e manual calculation is tedious. We use techniques of spatial statisti cs, directional data analysis and Bayesian image analysis to obtain th e orientation field and extract features from fingerprints. (C) 1997 P ublished by Elsevier Science B.V.