Page 128 - Automated Fingerprint Identification Systems (AFIS)

FROM PRINT TO IDENTIFICATION 113

the placement of the minutiae, ridge flow, core, and delta. If both examiners
agree on a decision, the result is considered final. If they agree that the candi-
date matches the inquiry, it is an ident. If they agree that the candidate is not
an ident, and no other candidate is produced, the subject is assigned a new SID
number and a rap sheet or criminal history is initiated. If the examiners differ
in their opinions, a supervisor or other examiner will be summoned. If the
images came from a livescan that is part of the booking process, the AFIS exam-
iners may request that the subject be re-rolled if still in custody. This may
produce a more clear set of prints from which to search. Another option is to
search other fingers on AFIS and/or to treat the images as latent print images
and search the latent cognizant database, if such a database is different from
the tenprint database.
For most tenprint searches, only one candidate will appear on the candidate
list. With improved coding, better matches, and faster throughput, the AFIS
systems have become better at producing an exact match in the first position
than in the past. Accuracy in image identification is a major selling point
for AFIS vendors. This accuracy is reflected in the AFIS candidate list. Ideally,
candidate lists will contain only the exact match, assuming the subject is
enrolled on the database. If the output list has more than one candidate, the
subject should appear in the first or second position of the list. No candidate
list should be produced when the subject has not been enrolled on the
database.
Tenprint searches are driven by a business model that requires a high degree
of accuracy, a sufficiently large database with high-quality enrolled images and
minutiae, and two or four good subject finger impressions to initiate the search.
There may be an administrative or legal requirement to make an identification
within a designated time period. For criminal processing, sometimes referred
to as Priority 1 processing, a fast turnaround is required to ensure the retrieval
of a rap sheet along with any wanted information or warrants prior to arraign-
ment. For civil applications, the issue of speed may not be quite as pressing,
but the search still must be concluded in a reasonable amount of time. Ten-
print processing is quantity driven, with perhaps hundreds of thousands or even
millions of transactions a year that must be completed accurately.
Some agencies are pursuing “lights out” tenprint searches. “Lights out”
replaces human decisions with mathematical probabilities. In the extreme
form, this means a tenprint search against the database with no human inten-
tion: no human who compares the candidates produced by the AFIS system; no
human to validate the hit/no hit decision. The response and rap sheet are auto-
matically sent to the inquiring agency. Agencies wishing to adopt the “lights
out” procedure may begin with a limited application on certain types of ten-
print searches, such as non-criminal AFIS searches that also match with the