BIOMETRICS IN LAW ENFORCEMENT AND CRIME PREVENTION

A Report to the Minnesota Legislature

Stephen Coleman

Center for Applied Research and Policy Analysis

School of Law Enforcement, Criminal Justice and Public Safety
Metropolitan State University
St. Paul, Minnesota

April 1999

1
BIOMETRICS IN LAW ENFORCEMENT AND CRIME PREVENTION

Executive Summary

This report is about the new computer technology of biometrics and its application to law
enforcement and crime prevention. The purpose of biometrics is to distinguish one person
from another based on each person's physical characteristics, such their fingerprints, face,
voice, eyes, or hands.

Minnesota's law enforcement agencies have successfully used biometric fingerprint

identification systems for over 20 years. This report describes these systems and others
that might be used in Minnesota. Some states, for example, are using biometric systems
to identify recipients of government benefits. These states are reaping substantial cost
savings by preventing people from receiving benefits under multiple identities. Other
states are using biometrics to enhance the security of drivers' licenses and stop people
from getting multiple licenses. In Britain, police agencies are using facial recognition
systems to automatically scan public surveillance TV cameras for known criminals.
Banks and credit card companies are testing biometric systems to make their credit and
ATM cards more secure and prevent the loss of millions of dollars from fraud.
Inexpensive biometric systems are now available that can prevent unauthorized access to
computers and computer networks.

Our review of biometrics technology shows that: It works very well; it can save money;
and it has good public acceptance. The wide variety of biometrics systems can present
difficult questions, however, for a government agency that wants to purchase one.
Different types of biometrics systems have different characteristics and levels of
accuracy, and none is compatible with any other. Before purchase, a system must be
carefully tested in a realistic setting.

Biometric systems in government and the private sector raise significant issues about
personal privacy, which the Legislature should address. Current laws provide few
safeguards for a person's biometric data. The potential value of biometrics for crime
prevention will not be realized if individual privacy is undermined. The report describes
what other jurisdictions are doing to safeguard privacy.

This report fulfills a request from the 1998 Minnesota Legislature for a study of
innovative uses of biometrics in law enforcement, the costs of potential uses, and a
discussion of related data privacy issues. This is one of several reports that the
Legislature commissioned in establishing the Center for Applied Research and Policy
Analysis at the School of Law Enforcement, Criminal Justice and Public Safety,
Metropolitan State University.

2
 CONTENTS

Executive Summary i

Biometric Systems 2

Fingerprints 4
Eyes 5
Hand Geometry 6
Voice 6
Face 7
Other biometrics 7
Accuracy 8

Applications 10

Criminal identification 10
Criminal background checks 12
Photo identification 13
Drivers' licenses 13
Corrections 15
Border control 17
Government benefit programs 19
Banking and credit cards 22
Computer security 23
Other applications 23
Public acceptance 24

Legality and Privacy 25

Guidelines for biometric privacy 26

Review and Conclusions 28

Notes 33

3
 TABLES

Table 1. Accuracy of several biometric systems as reported by manufacturers. 9

Table 2. Law enforcement applications of biometrics. 11

Table 3. State programs for authentication of welfare recipients. 21

Table 4. Approximate cost of generic biometric systems. 29

4
BIOMETRICS IN LAW ENFORCEMENT AND CRIME PREVENTION

A Report to the Minnesota Legislature

This report is about the new technology of biometrics and its application to law
enforcement and crime prevention. The purpose of biometrics is to distinguish one person
from another based on each person's physical characteristics. To do this, a computer can
use information about a person's fingerprints, face, voice, eyes, or hands.

The first application of biometrics in law enforcement was the automated identification of
fingerprints, and law enforcement continues to use biometric fingerprint data to solve
crimes, identify criminals, and keep criminal records. Increasingly, however, people will
be using biometrics in everyday life. Biometrics creates new opportunities for preventing
crime by accurately identifying people when they cash a check, collect welfare benefits,
use an ATM card, cross the border into the United States, sign on a computer network, or
enter a secure building.

The promise of biometrics may be tarnished however, if governments or businesses use it

to monitor and gather information on a person's private activities. People are just now
becoming aware of the threats to privacy inherent in this new role for biometrics.

In 1998 the Minnesota Legislature appropriated funds to Metropolitan State University to

establish the Center for Applied Research and Policy Analysis in the School of Law
Enforcement, Criminal Justice and Public Safety (Laws 1998, Chapter 367.) The
Legislature called for the Center to study:

5
 innovative uses of biometrics in law enforcement and evaluate the costs
associated with these potential uses . . . (and to) address any data privacy issues
that are raised by the use of biometrics in law enforcement.

In response to the Legislature, the report aims to answer these questions:

 What is biometrics, and how can Minnesota use it in law enforcement?

 How much do typical applications of biometrics cost?

 How accurate is biometrics at identifying people?

 Is biometrics legal?

 Is legislation needed to protect people's privacy from biometrics?

The study first reviews each type of biometric technology — fingerprints, voice, eyes,
hands, and face. We then investigate biometric systems used in other states, at the federal
level, and in other countries. Finally, we discuss laws and court rulings affecting
biometrics, and what other governments are doing about biometrics and privacy.

Biometric Systems

Biometric systems have two purposes: identification and authentication. Identification

means that the system is trying to find out who someone is. For example, when people
are arrested, their fingerprints are compared with thousands or millions of fingerprints on
file to determine if they have a criminal record. Identification also solves the problem of a
criminal who refuses to disclose a name or who gives an alias. Traditionally, highly
trained specialists did this fingerprint work manually. Now, the typical biometric
fingerprint system automatically reads a person's fingerprints with a video camera,

6
converts the fingerprint images into computer data, and searches the fingerprint file for
similar prints, which are stored digitally as electronic data.

A different situation arises when, for example, a person uses a credit card. The store clerk
wants to know if the person presenting the card is its owner. This is a case of
authentication or verification. The clerk may ask for additional proof of identity, such as
a driver's license. To use an ATM card, the owner must use a password or PIN number to
prove authenticity. Experience shows, however, that this type of security is weak, and
fraudulent use of cards, checks, and drivers' licenses costs businesses millions of dollars
each year.

In a biometric ATM system, a person's fingerprint data or another biometric could be put
on the magnetic strip on the back of an ATM card when the person first applies for the
card. To use the card, the owner would place it in a cash machine and put a finger on the
machine where the fingerprint would be read, converted into numbers, and compared
with the fingerprint data on the card. If the numbers match, the machine pays out. A
small computer in the ATM machine does the work. An alternative to a magnetic strip
holding the biometric data is a "smart card" that has a small computer chip memory, or
the biometric can be stored on a central computer and retrieved for comparison when a
person presents their identification card or password to the system.

An important difference between identification and authentication is the number of

comparisons the system makes to verify who the person is. For identification, the
computer may have to compare many thousands of fingerprints; for authentication, it
makes only one comparison — between the card and the person presenting it. A
biometric system that makes thousands of comparisons has to be much more accurate
than a system designed for a single comparison. The need for accuracy affects the choice
of a biometric system, because biometric systems vary in accuracy. The level of security
required is also an important concern. A nuclear plant, for instance, requires a very high
level of security, so it would need a biometric access system that made it almost
impossible to let in an unauthorized person.

7
Another important issue for a biometric system is how the public responds to it. Some
people may have an aversion to having their biometric identification recorded, or some
people may have trouble using a particular biometric system. These factors affect the
choice of a biometric system.

We will return to the accuracy and public response issues after describing the types of
biometric systems currently available.

Fingerprints

Each of our fingers has a pattern of lines or ridges that, together with breaks and forks in
the lines, are unique to each person. Fingerprints are well-established in law enforcement
and the courts as unique personal identifiers. No two people have ever been found to have
the same fingerprints.

Most biometric fingerprint systems work by finding the breaks and forks in the
fingerprint ridges, also called minutiae, and then converting information about their
position into a series of numbers, like coordinates on a map. Some systems use other
information about the ridges or pores of the skin. A dozen or more manufacturers
produce biometric fingerprint systems, but each company has its own mapping and
computational procedures, which are trade secrets. This makes their systems
incompatible with one another. That is, the data for a fingerprint read into one system
cannot be used on a computer of a different manufacturer. Compatibility among law
enforcement agencies that have different types of systems is achieved by re-creating a
visual image of the fingerprints from the computer data and sending the visual image to
another agency, where it is re-entered into their computer.

Law enforcement agencies take fingerprint data from all fingers. This greatly enhances
the accuracy of law enforcement systems over a commercial biometric system that might

8
use just one or two fingers for authentication. Fingerprint authentication may be
problematic for some persons if their occupation affects their fingerprints or because of
missing fingers.

Eyes

Two biometric systems use information about a person's eye; one looks at the pattern of
veins in the retina, the other uses the pattern of fibers, tissues, and rings in the iris. Both
of these biometrics are believed to be as unique as fingerprints, but their use in
identification lacks the established history of fingerprints.1

The retinal system was first to be invented. It is an intrusive system because it must shine
a light into the eye at close range to illuminate the retinal pattern; this also demands close
cooperation of the person being checked. Retinal systems are used mainly in very high
security institutions, such as nuclear plants, because they never let the wrong person in.

The iris-based system is relatively new but seems to be emerging as a favorite in the
market. The iris system requires only that a person look toward a video camera, which
can be several feet away. Eye glasses and contact lenses have no effect on the biometric,
nor does eye color. Similarly, change in size of the iris does not change the biometric. A
biometric system can use the fact that the iris adapts to light to verify that the eye is that
of a living person, not a photograph of an eye.

Some evidence suggests that patterns in the eye may change over time because of illness
or injury, but little is known about this. Blind people or people whose eyes have been
damaged by diabetes, for example, may not be able to use eye identification systems.

Only a small of number of companies sell eye identification systems. Almost all iris-
based systems use the same technology, which was invented by John Daugman, a British
scientist.2

9
Hand geometry

A hand has distinct characteristics in three dimensions, including the length, width, and
thickness of fingers, contour of fingers, veins, and other features, which are used in
biometric systems. Hand geometry systems are most common in building access control.
A person seeking admittance places a hand in a device that examines it with a video
camera and converts the video image to numbers in a computer for comparison with the
person's pre-recorded hand geometry.

Hand identification may be affected by changes in a person's fingers if they are swollen,
or by someone wearing a large ring. The system may also be sensitive to environmental
problems, such as temperature or light shining at the camera. People with paralysis in a
hand or tremors may not be able to use these systems.

Voice

People can be identified to a limited degree by their voice or speech pattern. To use a
voice recognition system, a person must pre-record specific words on the system. Later
when authentication is required, the system prompts the person to say one of these words.
A computer analyzes the speech pattern and tries to determine if the voice matches the
pre-recorded version. Notice that voice recognition is not the same as speech recognition,
in which a computer tries to "understand" what a person is saying.

Voice recognition suffers a variety of problems. A person's voice may change because of
illness or stress, and women are more difficult to identify than men. A noisy background
can also be a problem.

10
Face

Facial biometric systems are improving rapidly. 3 A video camera scans the face and
records the arrangement of features for comparison with previously recorded facial
images. How this is done is a proprietary secret of each manufacturer. One new system
uses the infrared heat pattern of the face as the biometric, which means the system works
in the dark.

Facial systems potentially have very wide application. They can identify digital
photographs of people stored on a computer, as when photos are made for drivers'
licenses in some states, and they can be used anywhere a video camera is available.
Facial recognition systems can automatically scan people's faces as they appear on
television or a closed-circuit TV monitoring a building or street.

Facial recognition can be impaired by people who change their appearance markedly, as
by growing beards, or who take on unusual facial expressions. The orientation of a
person's face toward the camera can also influence accuracy. These systems may not be
able to distinguish twins.

Facial systems are different from other biometric systems because they are the easiest to
use without a person's awareness, which raises some concerns about privacy. All other
biometric systems require a person's cooperation.

Other biometrics

A few biometric systems are available that use palm prints, which are similar to
fingerprint systems, and others use finger geometry, which is similar to hand geometry.
These biometrics are infrequently used, however, so we do not consider them further.
Some researchers are trying to develop biometric systems based on hand-written
signatures, personal odors, ears, sweat pores, the way a person types on a keyboard, or

11
body motions. To date, however, these are not sufficiently advanced for practical
application. DNA analysis is not usually considered a biometric because a person's DNA
cannot be read externally.

Accuracy

Biometric systems use two measures of accuracy, one for each type of mistake they
make. Suppose, for example, that a biometric system limits access to a building. One
mistake is to let in someone who doesn't belong in the building. The percentage of people
let in by mistake is the false acceptance rate (FAR). At other times the system might fail
to recognize a person who should be admitted. The frequency that this happens is the
false rejection rate (FRR).

The degree of error permissible in a biometric system is a decision that users have to
make in each application, taking into account the type of biometric system needed, cost,
security desired, and what happens if a mistake is made. Some biometric systems allow
these rates to be adjusted, which involves a trade-off between the two types of errors. If a
system reduces the chance of a person getting into a building by mistake, it increases the
chance that legitimate users will be denied admittance. So users must balance security
interests against the annoyance and problems that arise when legitimate users are denied
access. If too many bank customers are denied access at an ATM machine, for instance,
they may switch to a different bank. In some applications, high error rates are tolerable.
For example, when a fingerprint from a crime scene is searched against a fingerprint file,
a low chance of success in identifying the suspect is acceptable because there may be no
other possibility of finding the suspect, and a fingerprint examiner will always check the
computer's results for false matches before identifying a suspect.

In general, fingerprint and eye identification systems are the most accurate, followed by
hand geometry and face; voice recognition is least accurate. Table 1 shows typical
accuracy rates of biometric systems. These figures should be viewed cautiously, however,

12
as most information on accuracy comes from manufacturers; there is little independent
testing of biometric systems. A potential purchaser of a system should always test its
accuracy in a realistic situation before buying it. For example, a realistic trial of a bank-
by-phone system that used voice recognition in Switzerland found that accuracy was
disappointing and worse than the figures reported in experiments on voice recognition
(which are in Table 1).4

Table 1. Accuracy of several biometric systems as reported by manufacturers.

BIOMETRIC COMPANY FALSE FALSE

REJECTION ACCEPTANCE
RATE RATE
Retina EyeDentify 12.4% one-try virtually zero
0.4% 3-try
Iris IriScan 0.00076% 0.00076%
Fingerprint Identix, access 1% 0.0001%
control
Fingerprint Identix, computer 2% 0.001%
sign-on access
Hand geometry Recognition 0.2% one-try 0.2% one-try
Systems
Voice experimental 0.5%-1% 0.5%-1%
Face VisionSphere, 7.5% 1%
driver's license 4% 4%
photo, test 2.5% 15%
database
Face Identification 0.5% one-try 0.5% one-try
Technologies
International

Sources: Rayco Security (retina); Richard Wildes, "Iris Recognition: An Emerging Biometric Technology,"
Proceedings of the IEEE, 85 (1997):1361; Identix Corp., (fingerprint systems); Recognition Systems (ID3D
HandKey Reader); Fredric Bimbot, et al. "Speaker verification in the telephone network: research activities
in the CAVE project," Proceedings of the 5th International Conference on Speech Communication and
Technology, Rhodes, Greece, September 1997: 971-974; VisionSphere Technologies, Inc. (Canada),
UnMask+ test for West Virginia driver's license program with 30,000 photos by digital camera;
Identification Technologies International, PAC-1000 system specifications.

Some institutions are developing standards for accuracy. The European Union is
considering a standard for access control that would have a false acceptance rate of 1 per
100,000 (0.001%) combined with a false rejection rate of less than 1 percent.5 The British

13
banking community has called for a false acceptance rate of less than 1 per 10,000,000
(0.00001%).6

Accuracy of a system may also depend on how much information is extracted from the
biometric. For example, ten fingerprints are better than one. And a system that looks at
more features on a face or finger may be more accurate than one that uses fewer features.
Accuracy may also be improved by allowing a person to try using a system more than
once before rejection, or by combining several types of biometrics on a single system.

Applications

This review covers a wide variety of applications for biometrics technology.7 We include
law enforcement, corrections, drivers' licenses, government benefits, banking and credit
card fraud prevention, physical access control, and computer access control. Some of
these technologies are already being used in Minnesota, while others could be used
almost immediately. We also discuss applications that may not seem germane to
Minnesota, such as border control, but which show the effectiveness of different types of
systems. Unfortunately, few biometric systems have been rigorously evaluated, a
problem compounded by the rapid pace of change in the industry. We begin with
applications in traditional law enforcement activities, which are summarized in Table 2.

Criminal identification

Minnesota was the first state to install an automated biometric fingerprint identification
system. It identifies criminals on arrest and permits searching of prints found at crime
scenes.8 The Bureau of Criminal Apprehension (BCA) manages the system. For 20 years
the state has kept its place at the forefront of this evolving technology. Law enforcement
agencies throughout the state are connected by an electronic network that permits rapid
identification of criminals by their digitized fingerprints. The FBI has standardized the
electronic transmission of digitized computer fingerprints so they can be sent to the FBI

14
or between states for further identification. The latest development is the replacement of
traditional inked fingerprinting by "live scanning," which is the direct reading of
fingerprints by video camera and computer

Table 2. Law enforcement applications of biometrics.

Type Agencies Purpose

Fingerprint Most state identification bureaus, Identification of arrested persons
many police departments, FBI and latent prints from crime scenes.
INS Border control.
Retina Cook Co., Illinois, Sheriff's Office Identification of prisoners in jail.
Iris Lancaster Co., PA; Sarasota Co., Identification of prisoners in jail.
FL.
Face Los Angeles Co. Sheriff Searching mug shots.
Newham, England police Surveillance video cameras.
West Virginia Driver's licenses.
Illinois Driver's licenses.
Hand Federal and state prisons, 3 Prisoner identification.
Minnesota prisons
New York City corrections Probationer monitoring.
INS Border control.
Voice INS Border control.
Some corrections agencies Probationer monitoring.

The FBI is testing a new "NCIC 2000" system that will let law enforcement agencies
transmit digitized fingerprint information to the FBI to quickly identify whether a suspect
is on the national list of wanted persons. The sheriff's off ices of Hennepin and Ramsey
counties, the St. Paul police, and a local company, Digital Biometrics, plan to test a
portable fingerprint reader for NCIC 2000. The portable device will be used in a police
car or within a short distance from a police car to learn on the spot whether a suspect is
on the national wanted list. An image of a suspect's finger would be sent to the BCA and
on to the FBI. The initial test will place portable readers in 40 cars. These devices are
expected to cost about $2,500 to $4,000, with additional costs to upgrade police radio or
telephone systems to accommodate the fingerprint readers; federal funds are expected to
pay for the equipment.

15
It may take several years for police agencies to learn whether portable fingerprint readers
are useful and how many might be needed by a department. So we are unable to estimate
a cost for this biometric application.

Criminal background checks

Minnesota requires some classes of people to have criminal background checks as a

condition of employment or licensing. These groups include law enforcement, newly
licensed school teachers, school bus drivers, security guards, school district employees,
licensed day-care workers, foster-care providers, persons providing care to vulnerable
adults (as in nursing homes and hospitals), social workers, persons seeking liquor
licenses, building managers and caretakers, and persons working in the gambling
industry, among others.

BCA makes about 120,000 background checks each year. In addition, the Department of
Human Services has access to BCA's computerized criminal history file and makes about
100,000 checks each year. Private companies make another 65,000 yearly background
checks using public terminals at BCA. BCA usually does background checks by name
and date of birth only. If fingerprints are included with the request, BCA will also check
those to verify identity. Any background checks that go to the FBI must include
fingerprints.

In 1998 California began electronic submission of fingerprints through live scanning of

persons who need background checks; terminals are in sheriffs' offices throughout the
state. This service is expected to cut the time it takes to do a background check to 72
hours or less for 95 percent of applicants. Applicants pay a fee for the service. Alaska
will also be doing this. The cost of a terminal for live scanning is about $55,000.

16
We consulted BCA about whether a live-scanning terminal would help with their
background checks. They replied that it would be hard to incorporate live scanning into
their operations at this time. In the future, however, it might be a good option.

Photo identification

Facial recognition can be added to automated systems that capture and digitize mugshots
after people are arrested. In 1988 the Lakewood Division of the Los Angeles County
Sheriff's Department installed a system that can search the composite drawing of a
suspect or a video image of someone committing a crime against its database of digitized
mugshot photos. They also intend to use the system to search for suspects on "Megan's
Law CD," a photo database of registered sex offenders.9

British police are moving ahead with plans to automatically monitor closed-circuit
surveillance video cameras with facial recognition software. Britain has over 200,000
video cameras used for surveillance, many watching streets and shopping areas. In
Newham, a borough of London, with 140 street cameras and 11 mobile units, the local
police have planned a six-month trial of facial recognition at a cost of about $100,000.10
Video cameras will be monitored by a computer that will watch for known criminals, and
police will be alerted if they are seen on camera. The computer can compare 250 images
per second. The system will be tested at soccer games to scan the crowd for chronic
troublemakers. This technology can also monitor public video cameras for missing
children or scan for terrorists at an airport.

Drivers' licenses

Because a driver's license has become a key form of personal identification, its
authenticity is very important. Driver's license fraud contributes to other types of crime,
including passing bad checks, stolen identities, credit card fraud, and illegal purchase of
alcohol and tobacco by underage minors. States are taking more precautions to control

17
the issuance of a driver's license, and biometrics offers the best way to check a person's
identity.

West Virginia was the first state to apply facial recognition to a driver's license.11 When
people apply for a new license, their photo is stored digitally. Later, if someone applies to
replace a lost license or renew a license, their photo is again captured digitally and
automatically compared with their previous photo. The aim is to prevent people from
getting an extra license under a false identity. West Virginia gives people the option of
using a fingerprint for identification, which is done electronically. West Virginia has
installed 74 workstations at 54 locations to process drivers' licenses and expects to issue
450,000 each year.

Illinois will also be using facial recognition to help authenticate drivers' licenses.12 And
Illinois State Police will be able to access the system for a variety of law enforcement
purposes.

If a person's biometric data is put on a driver's license, it can be used to authenticate the
person at a store or other locations. First, the license is read like a credit card by swiping
it through a magnetic strip reader attached to a computer (a regular PC will do the job.)
Then, depending on the biometric, the person puts a finger on a fingerprint reader or
looks at a video camera so the computer can read the biometric a second time and
compare it with data from the license.

Several manufacturers sell biometric systems that run on conventional Windows-based

PCs and can be used at a store. The entire package — PC, biometric reader, and software
— can be purchased for $2,000 to $3,000; if a computer is already at hand, the biometric
technology might be added for as little as $300. We discuss PC-based biometrics again
when we review applications for computer security.

If a biometric is put on a driver's license, compatibility with other biometric systems

becomes a far-reaching issue. Compatibility with other government identification systems

18
might be desirable so that a license could be used for multiple purposes or to reduce
costs. Incompatibility, however, ensures a disconnection among government and
commercial identification systems, which might protect the privacy of license users. The
decision to put a biometric on a driver's license also involves many other considerations,
such as whether it should be optional, which biometric should be used, whether multiple
vendors are available (to provide competition), how information is stored on the license,
and so forth.

The Federal Highway Administration sponsored a study of biometric systems in the

licensing of commercial motor vehicle drivers. The study addressed two problems: the
verification of the person who has a license and the prevention of someone getting
multiple licenses. Federal authorities have been concerned that commercial drivers are
obtaining multiple licenses in different states. The National Biometric Test Center at San
Jose State University in California did the study, concluding that fingerprint identification
is the only feasible technology that can prevent drivers from getting multiple licenses.13
But systems using hand geometry, voice, or fingerprinting can do verification.

Corrections

Correctional institutions have been leaders in biometric systems. The Minnesota

Department of Corrections has hand geometry systems in the state's three medium-
security prisons (Faribault, Lino Lakes, and Willow River/Moose Lake). The systems
verify identities of staff, visitors, and volunteers to prevent an inmate from assuming their
identity and escaping. A person's hand biometric is put on a magnetic strip on an
identification card, which is read by the system and compared with the actual hand of the
person presenting the card before the person is allowed to leave the institution. Staff also
have photos on their identification cards to supplement hand geometry verification and
make false identification a remote possibility.

19
The department installed the systems in 1997 and continues to refine them, with an eye
toward putting similar systems in other prisons. The cost at Lino Lakes was $130,000,
which included a computer network and terminals at three access points to the prison.
The cost was greater than for an off-the-shelf system, however, because of computer
programming to tailor the system to the department's specifications. The system has the
added feature of continuously tracking which staff and visitors are in the prison.

The Minnesota Department of Corrections modeled its system after hand geometry
systems of the Federal Bureau of Prisons. In a number of federal prisons, visitors, staff,
and inmates have their hand biometric put on a photo identification card, which they must
carry with them. Card and hand readers are installed throughout the prison to control
access to various areas; the system also records everyone's movements. As reported to the
General Accounting Office by prison officials, one prison has made over 200,000 identity
checks with hand geometry without finding any errors.14 One system of this type with
eight hand readers, cameras, and related computer software cost about $90,000.

In 1990 the Cook County (Illinois) Sheriff's Office began using a retinal identification
system for prisoner identification and release.15 At that time the system cost about
$500,000 for 23 scanners and other equipment to connect the system, but prices for the
equipment have come down. It scans about 300 prisoners daily, comparing their retinal
biometrics against a database of more than 300,000. The system takes about 2 minutes to
identify someone in the database but only a few seconds to verify someone for access
control.

The Lancaster County (Pennsylvania) Prison installed a prisoner identification system

using iris-based technology in 1997. They paid about $10,000 for a stand-alone system.16
Sarasota County in Florida also plans to use this type of system in their detention center
to identify inmates prior to release.17 The center estimates it will process 18,000 inmates
annually.

20
In November 1998 Wisconsin contracted for a combined photo and facial recognition
system for the state's prisons, at 44 locations.18 It will cost $1.4 million.

In 1997 New York City contracted for an electronic probation monitoring system.19 The
system uses kiosks with video touch-screens that allow probationers and parolees to
check in electronically with probation officers on reporting dates. The system identifies
people by their hand geometry. (The manufacturer also makes a system that uses
fingerprints.) The city expects to use the system to monitor about 35,000 low-risk
probationers, allowing probation officers to spend more time supervising high-risk
probationers. New York paid $925,000 for their system.

A similar system was tested in Duluth by Arrowhead Regional Corrections on about 70

of 600 probationers beginning in 1994.20 The Legislature funded this $100,000 pilot
project.21 An evaluation by the Sociology Department at the University of Minnesota in
Duluth rated it well. The cost for a single kiosk in 1995 was about $50,500; additional
kiosks cost about $35,000.

Another biometric option for monitoring probationers is voice recognition. In one system,
a probationer is paged randomly and must call an 800 number that verifies who the caller
is.22 The location of the caller is checked by "Caller ID."

Border control

The federal Immigration and Naturalization Service (INS) has pioneered biometric
systems in border control. The number of people crossing into the United States daily is
enormous — 100,000 in El Paso, Texas alone — which makes prompt and accurate
identification necessary. In 1993, INS began a trial of its INSPASS control system, which
uses either hand geometry or, at one location, fingerprints.23 Business travelers who fly
into the United States at least three times yearly can apply for a biometric identification
card at an INSPASS enrollment center. Then, on arrival, the traveler puts the card into a

21
kiosk and puts a hand on the designated spot. If the system verifies the person's identity,
it opens a gate and records the process.

INSPASS started at New York's JFK terminal, Newark International Airport, and Pearson
International Airport in Toronto, and by 1997 over 60,000 people had enrolled. INS has
decided to add the program to at least six more airports. Other countries are developing
similar systems.

The U.S. Department of Justice evaluated the initial INSPASS system, which cost about
$3 million.24 Their audit found that INSPASS ". . . has the potential to be a cost effective
means of reducing processing time for frequent travelers . . . without sacrificing security."
The one-time cost of a kiosk — about $45,000 — is less than the on-going cost of a
border inspector, which is about $50,000 per year. And accuracy of the system is at least
as good as that provided by the normal manual inspection process. The audit also
recommended several improvements in kiosk design, enrollment accuracy, staff training,
the identity card, and the biometric. The auditors did a survey of travelers and reported a
high level of satisfaction among those who had used the system.

INS selected hand geometry for most of its kiosks because a person's hand biometric can
be put into a nine-character code that can be printed on any standard identity card in a
font that is compatible with the machine-readable optical standard used for passports.
INS tried one fingerprint kiosk, but fingerprint data takes more space and could not be
printed on the card, so the system had to store the data of all enrollees in a computer in
the kiosk. INS also had been reluctant to use fingerprints because of a presumed stigma
among travelers. The audit discovered, however, that 89 percent of travelers responding
to a survey had no objection to using their fingerprints. The auditors recommended
further testing of a fingerprint biometric in case the hand geometry system is not accurate
enough for nationwide implementation.

22
The auditors urged INS to consider using a "smart card," which has a computer chip that
can store much more data about the card-holder than a conventional magnetic card strip.
An INS identity card costs 40 cents; a smart card costs from $4.00 to $10.00.

INS has opened automated border crossing systems (PORTPASS) on the northern border
in Montana and in Coronach, Canada. These are remote locations that are not manned
around the clock. Here INS uses a voice recognition system to authenticate border
crossers because it is most able to withstand severe weather at these sites. To use the
system, a person lifts a phone, dials a special number on a keypad, then recites a specific
phrase. It a computer decides that the voice matches a pre-recorded version, it raises a
gate and admits the person.

On the Mexican border, INS is trying a facial biometric system to speed recognition of
daily commuters. As a vehicle drives through a special commuter lane, an electronic
device (a transponder) on the car sends information about the car to the system. The
system links the car with previously recorded information about the expected driver and
passengers who normally cross the border in the car, and the information is displayed on
a terminal for a custom's inspector, who can verify the car's occupants with a facial
recognition system. If they pass the check, they proceed across the border; if not, they are
sent to a conventional border inspection lane.

Government benefit programs

Biometrics are being used increasingly in government benefit programs for two purposes:
to secure the electronic transfer of funds to eligible recipients and to prevent "double
dipping," that is, to prevent a person from receiving multiple benefits under different
names.

In 1994 a federal task force recommended a substantial increase in electronic benefits

transfer (EBT) to welfare recipients. EBT uses personal identification cards similar to

23
credit cards to transfer funds to recipients through ATM machines; Minnesota law also
permits this.25 In 1995 the General Accounting Office (GAO) reviewed options for
increasing the security of EBT, focusing on the potential use of biometrics.26 GAO
concluded that "EBT alone does not effectively deter fraud in the delivery of food stamp
benefits. Thus, an EBT program without the enhanced security of biometric verification
raises a genuine concern about the potential for increased program costs and losses."
After reviewing various biometrics available, GAO reported, ". . . we believe fingerprint
verification is the biometric form that offers the greatest potential for success and
acceptance in securing EBT systems from fraud."

As of 1998 at least eight states had biometric systems for verifying eligibility of welfare
recipients; some also use them for EBT. Table 3 summarizes the state systems. Los
Angeles County started the first system in 1991 using fingerprint identification for its
550,000 welfare recipients. According to the GAO report, the county spent $9.6 million
to buy the system and had $14 million in savings between June 1991 and July 1994.
GAO also cited an evaluation of the county system by Ernst & Young showing that only
3 percent of clients had negative feelings about fingerprinting. The success of the system
led to its implementation in several other California counties, which share data on
recipients. California is moving ahead toward a statewide system.

On a smaller scale, Connecticut recently set up a statewide system that had about 150,000
welfare recipients enrolled as of February 1998.27 Their fingerprint-based system is
projected to cost $5.1 million over three years, which is about $6.64 for each person
enrolled of a prospective 200,000. The main effect of the system has been to reduce the
number of people who are registered for welfare; many previous recipients chose not to
enroll in the new system. Based on the smaller number of recipients, the state estimates
savings of $3.8 million annually for general assistance (GA) and $5.6 million or more for
AFDC. These savings do not include people who might be deterred in the future from
fraudulent applications for welfare.

24
Table 3. State programs for authentication of welfare recipients.

STATE BIOMETRIC BENEFIT SITES YEAR SCOPE ENROLLED COMMENTS

BEGAN
AZ fingerprints AFDC, food 89 1998 several in rollout
stamps counties
CA fingerprints; AFDC, GA, 1991 7 counties 625,000 going statewide
hand geometry food stamps
at one site
CT fingerprints AFDC, GA 70+ 1996 statewide 150,000 savings est. $9+
million annually
IL fingerprints or AFDC 1996 limited test of 17,000 recommended
retina both fingerprint over
biometrics retina
MA facial; or AFDC, GA, 42 in 1996 statewide 220,000 largest use of
fingerprints at food stamps 39 facial biometric
some sites offices
NJ fingerprints GA 24 1995 northern NJ 46,000
only
NY digital photo AFDC, GA, 207 at 1995 statewide 1 million, estimated $256
and signature starting food 58 going to 2 million savings
+ stamps sites million
fingerprints
TX fingerprints AFDC, food 1996 test areas 103,000 going statewide
stamps
NC fingerprints AFDC, food 1998 statewide starting up
and face stamps,
combined medical

Notes: (1) Data as of 4/30/98. (2) Titles of benefit programs may be different after welfare reform. (3) Most
states began with a pilot or limited program before statewide implementation. (4) NY and MA may also
use systems for electronic benefits transfer. (5) Fingerprint identification systems generally use two fingers.
(6) Several other states are planning systems.

Sources: Connecticut Department of Social Services, "The U.S. human services eight state biometric tour,"
http://www.dss.state.ct.us/digital/8_states/8states.htm; and "Report to the General Assembly, DSS digital
imaging project," March 1998.

Although other states have had good experiences with their biometric welfare systems, it
is difficult to extrapolate the potential benefits to Minnesota. Welfare reform and falling
welfare rolls might reduce the potential for cost savings. Another limiting factor is that
Minnesota would not be able to exchange data with nearby states because none have
these systems. In 1995 the Minnesota Legislature stipulated that identity verification
should be added to requests for proposal on the expansion of EBT.28 As in other states, a
biometric system could offer both reduced opportunity for welfare fraud and increased
security for recipients who get their benefits electronically.

25
Most states that have biometric welfare systems began with modest pilot programs. In
1996, for example, Texas started a one-year test of a fingerprint-based welfare
identification system in San Antonio at a cost of $300,000. The pilot system had 27
workstations at 14 sites and a database of about 100,000 welfare recipients. Texas has
since decided to expand to a statewide system.

Banking and credit cards

To this point we have discussed applications that are mainly governmental. These have
been the first large-scale uses of biometric technology. With the fall in prices for
biometric technology and the increasing power of personal computers (PCs) to do
biometric processing, businesses are turning to biometrics to prevent fraud in banking
and credit cards. Commercial applications of biometrics are pertinent to our study
because they are opportunities for crime prevention, and they also raise privacy concerns
that government may have to deal with.

MasterCard reported losses worldwide to fraud in 1997 that amounted to 7.7 cents for
every $100 of transactions, or a total loss of $462 million.29 The company believes that
biometrics will someday virtually eliminate fraud. To hasten that day, the company has
been testing a smart card with a fingerprint biometric, and in 1998 or 1999 will be
starting a pilot program with the card.

In Britain the Nationwide Building Society, which is the country's largest home mortgage
lender, began testing an ATM card in March 1998 that relies on the iris biometric instead
of a PIN number.30 To use the ATM, a customer puts an ATM card into the machine, and
a stereoscopic camera on the ATM machine photographs the customer's iris and checks
its pattern against the pre-recorded iris biometric associated with the cardholder located
in its main computer database or on the card itself. The trial was expected to last six
months, leading to a decision about nationwide implementation.

26
Computer security

Biometrics is developing rapidly for computer and Internet security. It can authenticate
the person signing on a computer or computer network and control access to sensitive
data over a network. Biometrics could guard government databases against snooping by
employees. Another application is to make financial transactions over the Internet more
secure. The biometric replaces passwords and PIN numbers. Law enforcement agencies,
the Department of Revenue, and other government departments that have confidential
data on networked computers may wish to consider adding biometric security. For
example, police agencies that share information on narcotics or gangs over a computer
network could limit access to this data more securely by a fingerprint than with password
control.

A variety of low-cost biometrics can be used with PCs. A small video camera can be
attached to a PC for about $100 to $300, depending on the computer's capabilities, and
some facial recognition software is under $100. Fingerprint readers for a PC are sold for
$300 or less. In 1998 Compaq began offering a fingerprint sign-on system for its PCs for
$99. Software for fingerprint authentication on a network or the Internet would be an
additional expense.

Other applications

A recent breakthrough in technology will further reduce prices and extend the range of
fingerprint verification.31 Most fingerprint scanners have a small video camera to capture
an image of the fingerprint. But fingerprint scanners will soon be available on a single
computer chip at an expected cost of $10 or less. A person will simply put their finger on
the chip itself to scan the fingerprint into the computer. Because the chip is so small and
thin, it can be mounted just about anywhere: on a computer keyboard, a doorknob, a car
ignition, or a cell phone. One firearms manufacturer is working on a gun that has a
fingerprint chip on the gun handle and a small computer inside the gun to authenticate a

27
fingerprint.32 The owner of the gun would touch the fingerprint sensor to activate the gun;
no one else could use it. The New Jersey legislature has been debating a bill that would
mandate a "child proof gun," which might use this technology.

Public acceptance

Most of today's biometric applications are restricted to small segments of the population.
Although biometrics have been well accepted in these cases, it is an open question as to
how the general public would react to biometrics in everyday life. Banks and credit card
companies are moving cautiously ahead with pilot studies to test public acceptance.

A national opinion survey of public attitudes on the use of fingerprints for personal
identification was done in 1996.33 The survey asked people about their personal
experience with fingerprinting, their awareness of identity fraud, their willingness to use
their fingerprint for identification, and the information policies they deemed important to
protect their privacy. The results showed that:

 55 percent of respondents had been fingerprinted and, of those, 87 percent

believed that it was appropriate.

 75 percent said they would be "comfortable" using an automated fingerprint

identification system.

 Over 75 percent approved the use of a fingerprint for check cashing or credit card
verification.

 83 percent did not agree with the statement that ". . . finger imaging treats people
like presumed criminals."

Over 90 percent of respondents supported policies that people should be informed in

advance of finger imaging, that they could find out the purpose of any request for identity
verification, and that their approval would be required before any organization could use
a stored fingerprint image for identification, unless required by law. Survey results were
generally consistent across all major demographic, regional, and occupational groups.

28
Legality and Privacy

Federal and state governments have used fingerprints to identify people for many years,
both in criminal cases and for civil purposes, such as drivers' licenses and criminal
background checks prior to employment or licensing. These purposes have consistently
withstood legal challenge. 34 Similarly, government can require photographs for
identification. Courts have not extended any constitutionally based privacy right to
people who are required to prove their identity with a fingerprint, although government
may have to show that it has a reasonable public purpose for taking a fingerprint. New
biometrics, as for the iris or hand geometry, have not been tested in court, but many
authorities believe that fingerprints have set a precedent.

Privacy issues emerge when a government agency does something with a person's
fingerprint that goes beyond the original purpose for its collection. California requires a
fingerprint for a driver's license. The California Supreme Court approved taking
fingerprints but ruled that indiscriminate dissemination of fingerprint records by the state
violated individual privacy rights.35

States that require fingerprints of welfare recipients have put different restrictions on
access to the fingerprint data.36 New York prohibits any use of fingerprint data beyond
the prevention of multiple enrollment; it cannot be used in a civil or criminal
investigation. Connecticut limits the use of fingerprint records to the welfare system and
to prosecution of welfare fraud. Pennsylvania, however, permits access to fingerprints by
the Pennsylvania State Police, the Board of Probate and Parole, and local law
enforcement authorities to find people who are wanted by the police on outstanding
warrants.37

People have much less protection of their biometrics in the private sector than in
government. Privacy in this context can be defined as the ability to maintain control over

29
the uses and dissemination of one's personal information.38 The greatest threat to privacy
is that companies might use people's biometrics to link up information on them, creating
a fairly detailed picture of a person's life, or get access to information without their
knowledge or consent. Courts have ruled that if people voluntarily turn over information
about themselves to others they generally should have no legal expectation of privacy.39
At the same time, however, biometric identification offers substantial protection against
stolen identities.

Facial recognition, especially when combined with public video surveillance, is a special
problem because people might not be aware that they have been identified by a remote
camera. This could be done via a live camera or by scanning video tapes. Indeed, people
out in public can be photographed or come under video surveillance without it being a
violation of their legal rights to privacy.

A parliamentary committee of the Canadian House of Commons that studied public video
surveillance raised many concerns about its relation to privacy.40 Some of the committee's
questions were: To what extent should video surveillance be done live versus taped?
Once a tape is made, who is the owner and who is entitled access? How should the
balancing of privacy rights with the benefits of new technologies be tackled in the area of
video surveillance? How should government handle the commercialization of personal
information derived from video surveillance? And so on.

Guidelines for biometric privacy

Governments around the world are actively working on guidelines and laws for the
privacy of biometric information. The provincial Information and Privacy Commissioner
of Ontario, Canada has been a leader. The office proposed a series of safeguards for
fingerprint identification records of the Ontario government in social assistance
programs.41 Recommended safeguards include:

30
  Encrypting the biometric, making it impossible for anyone to use it without
authorization.

 Restricting use of the encrypted finger scan to authentication of eligibility, so that

it cannot be used for surveillance or social control.

 Ensuring that an identifiable fingerprint cannot be reconstructed from the

encrypted biometric.

 Ensuring that the encrypted finger scan cannot be used to identify an individual in
the manner that a fingerprint can.

 Ensuring that a latent fingerprint from a crime scene cannot be matched to an

encrypted fingerprint in the database.

 Preventing unauthorized access to the biometric database by strict controls.

 Keeping personal information separate from the biometric.

Most of these elements became law in Ontario in 1997.42

In October 1998, the European Union put into effect a strong law restricting private-
sector use of personal information.43 The purpose is to bar companies from using
information about their customers in ways the customers never intended. The law bans
practices that are common in the U.S., such as selling information about customers for
marketing purposes. The law covers a wide range of information that companies collect
on customers, employees, and hospital patients, among others. It extends privacy
protections to information collected through credit cards, telephone records, magazine
subscriptions, and Internet usage. It includes the collection, recording, storage,
transmission, and erasure of personal data.

This law has the potential to disrupt electronic commerce between America and Europe
because no country in Europe will be allowed to transmit personal data to any country
that does not have comparable privacy protections. In contrast to Europe, the American
government has proposed that companies police themselves under self-regulatory
organizations. Negotiations between the U.S. and Europe are continuing on how to
resolve these very different philosophies on data privacy.

31
A clash of views on privacy and biometrics erupted in California in 1998 when the
California Senate debated a bill to limit private-sector use of biometrics.44 As introduced,
the bill would have prevented any person, business, or institution (including government)
from selling, transmitting, exchanging, or otherwise providing biometric identifiers to
third parties, with certain exceptions for law enforcement and welfare agencies.
Furthermore, biometrics and DNA data could not be used to discriminate against anyone.
The bill was amended in committee, however, to ban any requirement for a person's
biometric in the identification of persons cashing a check, or opening an account at a
financial institution. Financial institutions opposed the amended bill because they use
biometrics to verify non-customer payees, and because it would preclude facial scanning
in automated check cashing services.

Review and Conclusions

Our review of the characteristics of biometrics systems and their applications leads to
these conclusions:

 Biometric systems work very well.

Technology on the market now can successfully identify and authenticate people with a
high degree of accuracy. If an application involves hundreds of thousands of people or
more, however, fingerprints are still the best choice. Minnesota has 20 years of successful
experience with automated fingerprint identification. Iris-based systems may equal or
exceed fingerprints in accuracy, but the limited number of vendors and lack of precedent
for iris recognition make them less attractive. Hand geometry systems have proved
themselves in physical access control, as in prisons, where high levels of accuracy and
security are required. Voice recognition is least accurate but might be the best alternative
when identity must be verified over a phone. Facial recognition systems create
opportunities to identify people unobtrusively and without their cooperation, as in video

32
surveillance, and they can be added to digital photo systems used for mugshots or drivers'
licenses.

 Biometric systems can save money.

Biometric systems have proved their cost-effectiveness in state welfare systems and
border control. In general, the cost of a biometric system is less than the on-going cost of
hiring a person to do the same type of activity, such as building access control. Table 4
compares the typical costs of biometric applications. In criminal identification, biometric
technology can save investigators many hours of work and solve crimes that might not be
solved by traditional police practices.

Table 4. Approximate costs of generic biometric systems.

Application Cost
Single, stand-alone authentication system, $100- $3,000 for most biometrics
PC system, or network terminal $6,000 for iris-based system
Police car mobile fingerprint scanner $2,500 to $4,000
Single kiosk for authentication $50,000
Live scanning fingerprint terminal $55,000
Prison or building access control $10,000 for single terminal to $100,000 for
network of multiple terminals (hand
geometry or fingerprint readers); $500,000
for network of retinal readers.
Large kiosk network for probationers to $900,000
check in
Facial recognition and photo identification $1.4 million
system for prison system, multiple sites
State welfare identification $300,000 for a test site
$5 - $10 million for a large system

 Biometric identification has wide public acceptance.

33
Current applications and a national opinion survey show that people are usually willing to
use their fingerprints and hands to identify themselves. Public acceptance of biometrics
for banking and credit cards, however, is still being tested.

 Many different biometric systems are on the market.

A profusion of vendors are producing biometric technology. Usually, it is good to have a

wide choice among biometric systems and companies that sell them. It is questionable,
however, how many companies will survive the competition or be able to keep pace.
Before selecting a vendor for a biometric system, a government agency should consider
its needs for long-term maintenance and its ability to update the technology as it evolves.

 Biometric systems have little compatibility with one another.

Not only are different biometrics incompatible, most vendors have unique, proprietary
technology in their products even if they use the same biometric. Government agencies
may find advantages in adopting compatible systems, but compatibility also makes it
possible to share data on people across systems, which might infringe on people's
privacy.

 Biometrics systems should be tested thoroughly before purchase or before

expanding to larger systems.

Buyers need to make their own assessment about the accuracy and reliability of biometric
systems and not rely totally on claims of manufacturers. A system that works well for a
small number of people may not be satisfactory if thousands of people need to be
enrolled in the system. A system must be tested for its ability to reject counterfeit
biometrics. Human factors in a real-life setting can also have a great impact on the
viability of this technology. Most biometric systems need the cooperation of users. If a
system produces too many false rejections, or is difficult to use, it will not be successful.
Some people may have problems with their biometric being read; they must be

34
accommodated. For example, welfare recipients cannot be denied benefits because a
machine did not read their biometric correctly. Environmental conditions can also affect
the working of a biometric system.

 In many applications, one must consider the appropriate identification

card technology.

Except for criminal identification, most uses of biometrics call for a person's biometric to
be put on an identification card. If a person's biometric is not put on a card, the system
must be able to retrieve the person's pre-recorded biometric from a central computer
database. Typically, cards use magnetic strips to hold data, but "smart cards" that have a
built-in computer chip may be a better option for a modest increase in cost. They can
hold more biometric data, as well as digitized photos and other useful information, and
they can be updated easily. Other types of cards are also coming on the market that
provide more security for the data. A study of biometrics for the European Commission
recommended that for the greatest security and to enhance individual privacy, cards
should be used to store biometrics whenever possible, rather than putting the biometric in
a central computer database.45

 Biometric privacy laws are needed.

People are generally willing to use a biometric for identification, but many are concerned
about what might happen to data that is collected about them. Biometrics increase the
potential for abuse of personal privacy by making it easier for governments and
businesses to gather and connect information about a person. Video surveillance and face
recognition are particular concerns, because people may not be aware of being on
camera. States that use a biometric to verify recipients of government benefits have legal
safeguards against improper use, but existing laws put few other restrictions on the
collection or use of biometrics by governments or the private sector. While Canada and
Europe have been strengthening privacy rights in law, the U.S. has taken the path of self-
regulation by private organizations. The potential of biometric technology to protect

35
identities, solve crimes, safeguard computer databases and networks, and deter fraud will
not be realized, however, if people's privacy is compromised.

36
37
1

NOTES

Richard Wildes, "Iris recognition: An emerging biometric technology," Proceedings of the IEEE, 85 (September 1997):
1348-1363. Wiles states (p. 1349) that ". . . large-scale studies that specifically address the distinctiveness and stability of
the iris, especially as a biometric, have yet to be performed."
2
John Daugman, "The scientific basis for iris recognition — Recognizing persons by their iris pattern," IriScan Corp.,
http://www.iriscan.com/basis.htm
3
Juan Velasco, "Teaching the computer to recognize a friendly face," New York Times, October 15, 1998, p. D7.
4
"Report D 7.1 - CAVE SV Result and analysis of fieldtests in banking applications," http://www.ptt-
telecom.nl/cave/download.html. Research on speaker verification (SV) technology has been heavily supported by the
European Union and Swiss government as part of the CAVE (CAller VErification in Banking and Telecommunications)
and PICASSO projects.
5
Despina Polemi, "Final Report — Biometric techniques: Review and evaluation of biometric techniques for identification
and authentication, including an appraisal of the areas where they are most applicable," European Commission, April 1997:
17; http://www.cordis.lu/infosec/src/stud5fr.htm.
6
Polemi: 9.
7
A good source of information and links on all types of biometric systems is the Biometric Consortium,
http://www.biometrics.org/. See also the new book edited by Anil Jain, Ruud Bolle, and Sharath Pankanti, Biometrics,
Personal Indentification in Networked Society, Boston: Kluwer, 1999.
8
Stephen Coleman, "Minnesota's automated latent fingerprint identification system." Proceedings of the Fourth
International SEARCH Symposium, Washington, D.C., 1979.
9
"Los Angeles Sheriff's Department installs facial recognition software," press release from Viisage Technology, Inc.,
http://www.viisage.com.
10
"Candid camera for criminals," BBC News, Tuesday, October 13, 1998 at 18:48 GMT. See
http://news.bbc.co.uk/hi/english/uk/ ; and Robert Matthews, "Security camera can find a face in crowds," Electronic
Telegraph, UK News, Sunday, 6 July 1997; http://www.telegraph.co.uk:80/.
11
"West Virginia becomes first state to issue driver's licenses using facial recognition technology," press release from
Polaroid, Corp., March 24, 1998; http://digitalimage.polaroid.com/polinfo/press-releases/march98/032598a.html.
12
"Viisage technology, Inc. to implement facial recognition system for Illinois Secretary of State," press release, February
26, 1998; http://www.viisage.com/illfr.htm.
13
San Jose State University, "The Federal Highway Administration conducting biometrics standards development study,"
June 1997, http://www.engr.sjsu.edu/biometrics/fhwa.html.
14
General Accounting Office, "Electronic benefits transfer: Use of biometrics to deter fraud in the nationwide EBT
program," September 1995 (GAO/OSI-95-20).
15
GAO report.
16
Keith W. Strandgerg, "Biometric ID," Corrections Forum, 1997, http:www.prisons.com/cforum/id.html.
17
"Sarasota County detention center adopts IriScan's iris recognition technology," press release from IriScan, Inc.,
http://www.iriscan.com/sarasota.htm.
18
"Wisconsin department of corrections chooses Viisage," press release from Viisage Technology, Inc., November 18,
1998.
19
"New York City awards $925,000 contract to Pacer Infotek for electronic probation management systems," press release,
March 5, 1997; http://www.corrections.com/products/news/crad.htm.
20
"Kiosk check-in for probationers," Government Technology, May 1995, http://www.govtech.net/.
21
Laws 1993, Ch. 146, Art. 2, Sec. 28, Subd. 1. An evaluation report was also required by July 1, 1994.
22
"VoiceTrack," http://www.voicetrack.com/
23
U.S. Department of Transportation, "Volpe engineers use biometrics to help ease border crush," Volpe Transportation
Journal, Spring 1997; http://web1.volpe.dot.gov/pubs/journal/spring97/biomet.html.
24
U.S. Department of Justice, "Immigration and naturalization service passenger accelerated service system pilot program,"
Office of the Inspector General, Audit Division, Audit report 95-8 (March 1995);
http://www.usdoj.gov/oig/inspass1/inspass1.htm.
25
M.S. 256J.39, 256D.09, 256D.47.
26
GAO report.
27
Connecticut Department of Social Services, "Report to the General Assembly, DSS digital imaging project — An
overview of the DSS digital imaging project implementation and first year results," March 1998.
28
M.S. 256.9850.
29
"MasterCard wins another round against fraud," press release from MasterCard, May 11, 1998,
http://www.mastercard.com/.
30
"World's first PIN-less ATM hits Britain's High Street today," press release from NCR, April 23, 1998,
http://www3.ncr.com/.
31
Miquel Helft, "Digital fingerprints: A key of the future," Wired News, September 30, 1998; http://www.wired.com/.
32
David Kocieniewski, "Trenton debates requiring guns only owner can fire," New York Times, September 24, 1998.
33
Alan F. Westin, "Public attitudes toward the use of finger imaging technology for personal identification in commercial
and government programs," National Registry, Inc., August 1996.
34
John D. Woodward, "Biometric scanning, law & policy: Identifying the concerns — Drafting the biometric blueprint,"
University of Pittsburgh Law Review, 59 U. Pitt. L. Rev., Fall 1997. The Supreme Court has also ruled in United States v.
Dionisio (93 S. Ct. 764) that a grand jury witness can be compelled to furnish a voice exemplar, as it does not violate the
Fifth Amendment privilege against self-incrimination, and further that a voice exemplar is not protected by the Fourth
Amendment prohibition against "unreasonable searches" because the voice is a physical characteristic constantly exposed to
the public. In yet another important case, Whalen v. Roe (97 S. Ct. 869), the Supreme Court ruled that a state could collect
the identities of medical patients who got prescriptions for certain drugs: The patient identification requirement was the
product of rational legislative action and was a reasonable exercise of state police powers; it did not impair any privacy
interest protected by the Constitution.
35
Perkey v. Department of Motor Vehicles, 42 Cal. 3d 185 (1986).
36
Ann Cavoukian, "Privacy and biometrics: An oxymoron or time to take a 2nd look?" Office of the Information and Privacy
Commissioner, Ontario, Canada; http://www.ipc.on.ca/web_site.eng/matters/sum_pap/papers/cfp98.htm.
37
Woodward; Pennsylvania Act of June 30, 1995, ch. 71, 1995 Pa. Laws 20.
38
Cavoukian.
39
Smith v. Maryland, 99 S. Ct. 2577 (1979).
40
The Honorable Sheila Finestone, Chair, "Privacy: Where do we draw the line?" Report of the House of Commons
Standing Committee on Human Rights and the Status of Persons with Disabilities, April 1997.
http://www.parl.gc.ca/committees352/huso/reports/.
41
Cavoukian.
42
Statutes of Ontario, Bill 142, Chapter 25, 1997.
43
Edmund L. Andrews, "European law aims to protect privacy of data," New York Times, October 26, 1998, p.1. Fiona M.
Carlin, "EU to install data privacy standards," National Law Journal, October 27, 1997.
44
California Senate, SB 1622 (Peace), as amended July 6, 1998.
45
Polemi: 36.