Fingerprint Sample & Directional Image
For the type classification algorithm, some sample results are presented below. For the clear image (top row of figure) shown, the correct number of core-delta points and their locations were found. However, for the unclear fingerprint (bottom row of figure), just one delta point was found. We think the other point was not found because of the large black smudge of ink in that portion of the fingerprint. Based on the clear sample, however, it appears that the algorithm successfully computes the directional image and the core-delta points for a good sample fingerprint. Also, looking at location and points estimates for other unclear samples, the same trend appears. The system finds the core-delta points when that portion of the image is clear.
Fingerprint Sample & Directional Image
For the correlation algorithm, we computed correlation values for a sample of our fingerprint database. However, we did not use a preliminary type classification to limit the comparisons due to unreliable results from the classification algorithm because of unclear fingerprints. We performed the correlation algorithm for an input sample size of 25 fingerprints and for a database of 20 fingerprints. Of the 25 fingerprints which were used as inputs to the system, 20 were from people who were also in the database. Of these 20 trials, 9 matched correctly while 11 matched incorrectly. Of the 5 samples which were not in the database, 2 were listed as no match and 3 were listed as matches. For the matches in both cases, the highest correlation value was not necessarily clearly separated from the other correlation values.