Page 102 - Automated Fingerprint Identification Systems (AFIS)

AFIS SUMMAR Y—HOW THE SYSTEM WORKS 87

Table 4.1
Before AFIS AFIS AFIS Changed the
Identification Business
Model
Finger classification Coder identifies minutiae
Fingerprint cards Images on RAID storage
Magnifying glass High-resolution monitors
Manual or semiautomated search Fully automated search
Photo, mail, laser fax Livescan
Response in hours, days Images on RAID Storage
Response in minutes

is usually the result of careless or inattentive image capture by the booking
officer or technician, resulting in reduced opportunities for identification.
Many forensic databases are populated with inked tenprint records that were
converted at a pixel resolution of 500 pixels per inch (ppi). Newer conversions,
at 1,000ppi or higher, provide more definition and extract more minutiae,
even from poor-quality images. If the local database has images at 500ppi and
the AFIS that made the identification has images at 1,000ppi, the latter has
more information to work with and thus a better chance of making the
identification.
When subjects were fingerprinted using the ink and roll method, three sets
of prints were taken. One card was retained for the local agency, one card was
sent to the state identification bureau, and one card was sent to the FBI. The
quality of the three cards could be vastly different, which could affect the results
of the search. With livescan systems increasingly replacing inked and rolled ten-
print cards, however, the subjects are rolled only once. The images that are sent
to the state AFIS, IAFIS, and kept in the local repository are all identical.
Occasionally missed identifications are caused by a clerical error made by
the booking officer on a critical piece of data, such as the sex of the subject.
Because AFIS searches only those records matching the given parameters, this
type of error would eliminate the subject from the search entirely. There are
other explanations as well, including the software used for matching and coding
the minutiae. While each vendor claims that their software is superior, there
are differences in their coding and matching algorithms. While unlikely, a poor-
quality record on the database might be identified by one vendor but not
another. This is more of an exception than the rule.
Yet another consideration is the version of software used in various compo-
nents of the AFIS system. AFIS systems that were installed 10 years ago proba-
bly had the records converted using software that was state-of-the-art at the time,
but that has now been replaced with better, more accurate software. Some agen-
cies have reconverted their entire databases using the newer coders and have