Page 91 - Automated Fingerprint Identification Systems (AFIS)

76 AUTOMATED FINGERPRINT IDENTIFICATION SYSTEMS

and civil databases, and use a “binning” structure. The “binning” allows for a
more rapid search of a database by limiting the search to only one segment of
the database at a time, e.g., males with a whorl pattern on finger number 2.
Each system contains a minutiae database of these images. It is this minutiae
database that is actually searched by the AFIS matcher. It includes finger image
characteristics such as the minutiae location, ridge flow direction, and distance
between minutiae points. Some systems also include ridge information. The
extracted minutiae, direction of ridge flow, etc., of the submitted finger images
are compared or matched against the records in the database of minutiae
records.
Large state and federal agencies receive thousands of search requests every
day. Some of these inquiries are for criminal processing as the result of an
arrest; others are for civil processing, such as an application for a job, permit,
or license. All of these submitted records are searched against the tenprint data-
base. If a match occurs, the inquirer is sent the SID number and the criminal
history. If there is no record matching the images in the tenprint file, a new
SID number is assigned, the images are added to that file, and the inquirer is
so advised. If the submission was related to a criminal inquiry, the images would
also be uploaded to the latent cognizant (criminal) file.
In addition to the database used for known civil applicants and the criminal
file, law enforcement agencies using AFIS also have an unsolved latent file. The
unsolved file contains records from criminal cases for which no identification
has been made following a latent print search. The expectation is that either
the individual has never been enrolled in the AFIS database and consequently
could not be identified, or the individual is in the database, but because of the
low quality of the tenprint record and/or the latent prints, no identification
could be made. At some time in the future, the image/minutiae on the data-
base may be updated and a tenprint to unsolved latent search may be initiated
with better success in making an identification. Or the matchers might be
upgraded, resulting in increased accuracy in selecting candidates following a
search.

4.2 PROCESSING OVERVIEW

4.2.1 TENPRINT

Consider the following simplified generic description of an arrest identification
process. The process begins at a local police agency when an individual is
arrested. The appropriate arrest information is entered into the local agency’s
booking system; fingerprints are taken by ink and roll or, more increasingly,
are electronically captured on FBI-certified equipment. The proper finger