Page 262 - Automated Fingerprint Identification Systems (AFIS)

CASE STUDY—DIAMONDS IN THE ROUGH 247

and latent cog databases, conducted by the Manager of Tenprint Operations,
Michael Tymeson, showed improvements in accuracy and performance. The
investment in funding and personnel resources was reaping huge benefits. Man-
agers began to ask if other opportunities for improvements existed.

10.5 OPPORTUNITIES FOR INCREASING
UL FILE IDENTIFICATIONS
Since the inception of SAFIS operation in 1989, latent print examiners had
been saving their unsolved cases to the UL file. A field in the case file, the UL
retention year, allowed the examiner to have the case automatically removed
after a certain date. For example, if a burglary occurred in 1991, the statute of
limitations was normally 7 years, i.e., 1998. In 1999, the case would be admin-
istratively removed from the UL file if the examiner so desired. This process
allowed the examiner to spend time only on cases of value and eliminated
unnecessary time commitment for cases that might never be prosecuted. Some
cases, such as homicides, have no statute of limitations and would be retained
in the UL file in perpetuity.
The introduction of new coders, new matchers, and a recoded latent cog
database demonstrated immediate improvements in the number of latent print
identifications that were made. The new imaging technology could mask back-
grounds, the coders could identify minutiae more exactly, and the matchers
searching the recoded database presented better candidates with higher match-
ing scores. The improvements in latent to tenprint (LT/TPlc) searches led to
the question, “Could more cases on the unsolved latent file be solved?”
The DCJS management team recognized that the unsolved latent file con-
tained three groups of records: (1) those that had been created using the
original coders and matchers, (2) those created during a transition when new
matchers and coders were in place, but the database was primarily composed
of originally coded records, and (3) those that had been created since July of
1999, when the new coders and matchers were introduced. These older cases
searched on a database whose minutiae features were extracted with an earlier
version of coder technology. Now the entire database had been reconverted,
and more and/or better minutiae were available. Table 10.1 shows these com-
binations as the system transitioned from the “original” SAFIS platform to the
new SAFIS platform.
With approximately 100,000 latent print images on the UL file, the oppor-
tunity for making identifications on older cases was obvious. DCJS staff, with
the assistance of Sagem Morpho staff, searched the UL file for cases entered
by the latent print examiners, and created two lists for each examiner that
included each case number, the date of entry, the original crime (e.g.,