21.02.2013 Views

Advances in Fingerprint Technology.pdf

Advances in Fingerprint Technology.pdf

Advances in Fingerprint Technology.pdf

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

f<strong>in</strong>gerpr<strong>in</strong>t was isolated. This area was chosen as it was the average area found<br />

<strong>in</strong> a set of 300 latent f<strong>in</strong>gerpr<strong>in</strong>ts. The simulated latent pr<strong>in</strong>ts, each actually<br />

a subset of the orig<strong>in</strong>al electronic f<strong>in</strong>gerpr<strong>in</strong>t record, were then compared<br />

aga<strong>in</strong>st each of the 50,000 full f<strong>in</strong>gerpr<strong>in</strong>t records.<br />

The same software was used to compare f<strong>in</strong>gerpr<strong>in</strong>ts, compute scores,<br />

and consolidate the scores. The same statistical analysis was performed.<br />

As with the first experiment, exam<strong>in</strong>ation of the Z scores from the nonidentical<br />

pr<strong>in</strong>ts revealed a number of unusually high scores. These <strong>in</strong>cluded<br />

the four same-f<strong>in</strong>ger, different pr<strong>in</strong>t comparisons discovered <strong>in</strong> the first<br />

experiment, here show<strong>in</strong>g Z scores of 8.00, 7.58, 8.01, and 8.38. Additionally,<br />

another same-f<strong>in</strong>ger, different pr<strong>in</strong>t pair was discovered, show<strong>in</strong>g Z scores of<br />

3.91 and 5.85.<br />

Similar to the first experiment, these different pr<strong>in</strong>ts from the same f<strong>in</strong>ger<br />

were excluded from the analysis. Meagher, Budowle, and Ziesig proceeded<br />

by group<strong>in</strong>g the Z score results accord<strong>in</strong>g to the number of m<strong>in</strong>utiae <strong>in</strong> the<br />

simulated latent pr<strong>in</strong>t and a chart was prepared of Z score vs. m<strong>in</strong>utia count.<br />

The highest Z scores for the non-identical pr<strong>in</strong>t comparisons were plotted,<br />

along with the Z scores for the comparisons of the simulated latent pr<strong>in</strong>ts<br />

with their correspond<strong>in</strong>g orig<strong>in</strong>al electronic record (under the hypothesis of<br />

equal means).<br />

Meagher, Budowle, and Ziesig concluded by calculat<strong>in</strong>g probabilities that<br />

simulated latent pr<strong>in</strong>ts with a particular m<strong>in</strong>utia count would be identical<br />

to a subset of any particular f<strong>in</strong>gerpr<strong>in</strong>t. For a simulated latent with a small<br />

number of m<strong>in</strong>utiae (c.f. 4) they stated that the probability was less than<br />

1/10 27 , decreas<strong>in</strong>g with larger numbers of m<strong>in</strong>utiae (c.f. 18) to 1/10 97 (the<br />

same probability offered <strong>in</strong> the first experiment for complete rolled f<strong>in</strong>gerpr<strong>in</strong>ts).<br />

Follow<strong>in</strong>g their earlier analysis, Meagher, Budowle, and Ziesig compared<br />

the larger (4 m<strong>in</strong>utia) probability with the world population, conclud<strong>in</strong>g that<br />

the chance of any two (small) m<strong>in</strong>utia subsets from any (different) f<strong>in</strong>gerpr<strong>in</strong>ts<br />

on earth be<strong>in</strong>g identical was 59/10 18 .<br />

Discussion of the Meagher, Budowle, and Ziesig Model<br />

The Meagher, Budowle, and Ziesig model has been discussed by Stoney 62-64<br />

and by Wayman. 65 The major issue with this model is that, with the exception<br />

of the excluded data, the Meagher, Budowle, and Ziesig experiments do not<br />

<strong>in</strong>clude compar<strong>in</strong>g two different pr<strong>in</strong>ts of the same f<strong>in</strong>ger with each other.<br />

All actual f<strong>in</strong>gerpr<strong>in</strong>t comparisons have this aspect, and every f<strong>in</strong>gerpr<strong>in</strong>t<br />

exam<strong>in</strong>er knows that no two impressions from the same f<strong>in</strong>ger are exactly<br />

alike. There would never be an occasion to compare a s<strong>in</strong>gle f<strong>in</strong>gerpr<strong>in</strong>t with<br />

itself.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!