Benefits of biometrics vs


23 Φεβ 2014 (πριν από 7 χρόνια και 6 μήνες)

248 εμφανίσεις


Why biometrics

Depending on the applications, the benefits of

using or deploying biometrics may be increase security
increased convenience,reduced fraud, or delivery of enhanced services.

Regardless of the rationale for deploying

ics, these are two common elements.

1) the benefits of biometric usage and deployment

are derived from having a high degree of certainty
regarding an individual's identity

2)The benefits lead directly or indirectly;y to cost

saving or to reduced risk of

financial losses for and
individual or institution

Benefits of biometrics vs

traditional authentication methods

Increased security

Increased convenience

Increased accountability

verification and identification.


systems answer the question
, “Am I who I claim to be?” by requiring that a user

claim an identity in order for a biometric comparison to be performed. After a user claims an identity,
he or she provides biometric data, which is then compared against his or her enrolled biometric da

Depending on the type of biometric system, the identity that a user claims might be a windows user
name, a given name or an ID number; the answer returned by the system is match or no match.
verification is often referred to as 1:1(one
one). The pr
ocess of providing a user name and biometric
data is referred to as authentication.

Identification systems answer the question, “who am I?” and do not require that a user claim an identity
before biometric comparisons take place. Identification is often re
ferred to as 1:N(one
to N or one
many), because a person's biometric information is compared against multiple(N) records.

How Biometric matching works

A user initially enrolls in biometric systems by providing biometric data, whi
ch is converted
into a template.

Templates are stored in biometric systems for the purpose of subsequent compariosn

In order to be verified or identified after enrollment, the user provides biometric data, which is
converted into a template.

The verificati
on template is compared with one or more enrollment templates.

The result of a comparison between biometric templates is rendered as a score or confidence
level, which is compared to a threshold used for a specific technology system, user, or

If the score exceeds the threshold, the comparison is match, and that result is transmitted

if the score does not meet the threshold , the comparison is not a match and that result is

Key terms and process involved in enrollment and te
mplate creation


the process by which a user's biometric data is initially acquired, assessed processed and stored in the
form of a template for ongoing use in a biometric system is called enrollment. Quality enrollment is a
critical factor in t
he long
term accuracy of biometric systems. Low
quality enrollments may lead to
high error rates, including false match rate and false non match rate.


Is the process by which a user provides biometric data to anaccquistion device
the hardwa
re used to
collect bbiometrcu data. Users must be cognizant of the manner in which they present biometric data in
order to be verified and identified successfully.

Biometric data.

The biometric data users provide is an unprocessed image or recording of a

characteristic. This unprocessed data is also referred to as raw biometric data or as a biometric sample.
Depending on the biometric system, a user may need to present biometric data several times in order to

Once biometric data has been acquire
d, biometric templates can be created by a process of feature

Feature extraction

The automated process of locating and encoding distinctive characteristics from biometric data in order
to generate a template is called feature extraction.

curacy in Biometric systems

To developing effective matching algorithms,
biometric companies have primarily concerned
themselves with performance in highly controlled environments.

Companies often assess their technology's
accuracy more specifically, the
accuracy of their matching
algorithms by using static or artificially generated templates,images and data.

The key performance metrics in biometrics are
false match rate, false non match rate and failure to
enroll rate.

False Match rate

A biometric solu
tion's false match rate is the
probability that a user's template will be incorrectly
judged to be a match for a different user's template.

False Non match rate

Is the probability that a user's template will be
incorrectly judged to not match his or her


enroll rate

A system’s failure
enroll rate represents the

probability that a given user will be unable to enroll in
a biometric system.

Derived metrics

There are two metrics used to reflect the overall


capabilities of a biometric technology.

These derived metrics are generated from analysis

of FMR,FNMR and FTE.

1)Equal error rate

verify(ATV) rate

Equal error rate

A more valuable derived metric is the ability

verify rate.ATV is a combin
ation of failure

and false nonmatch rates and indicates the overall percentage of users who will be capable of
authenticating on a daily basis.


Layered Biometrics

Layered biometric solutions are those that require

the sub
mission of more than one biometric
characteristic for verification such as finger
scan and voice
scan or finger
scan and facial scan.

There are two types of layered biometrics:

1) parallel


(diagram refer page 133 figure9.2)


Finger scan components

scan systems comprise image acquistioin

hardware, image processing components, template
generation and matching components, and storage components.

The surface on which the finger is placed is call

a platen.

A platen is one piece of a finger
scan module, the

basic building block of a peripheral or standalone
scan device.

How finger scan technology works

There are five stages involved in finger


verification and identification :

Finger print image acquistion

Image processing

Location of distinctive characteristics

Template creation

Template matching.

Image acquisition

The first challenge facing a finger
scan system is

to acquire a high
quality image of the fingerprint.
quality is measured in dots per inch(DPI) more DPI means a higher resolution image.

Image processing

Once a high quality image is acquired, it must be

converted to a usable format.Image processing
subroutines eliminate gray areas from the image by
converting the fingerprint image's gray pixels to
white and black,depending on their pitch.

Location of

The fingerprint comprises ridges and valleys that

form distinctive patterns, such as swirls ,loops, and

int ridges and valleys are characterized by

discontinuities and irreularities known as minutiae
these are the distinctive features on which most finger
scan technologies are based.

Template creation

Vendors utilize proprietary algorithms to map

ngerprint minutiae. Information used when mapping
minutiae can include the location and angle of a minutia point, the type and quality of minutiae and
the distance and position of minutiae relative to the core.

Template matching

Comparing enrollment
and verification templates

does not result in an exact match. Compare the
generated score with threshold. If it is higher than the threshold then the matching is success
otherwise failed.

Competing Finger
scan Technologies

1)optical technology

con technology's

3)ultrasound technology

The user places a finger on a coated platen, built of hard coated plastic or coated glass.

Optical technology has several strengths: It has been proven reliable over time is resistamt to
electrostatic discharge
is faily enexpensive, and can provide resolution up to 500 DPI, the benchmark
for high
quality fingerprint images.

Weaknesses include size( the platen must be of sufficeint surface area and depth to capture quality
images), a sporadic tendency to show late
n prints as actaul finger prints, and a susceptibility to fake

Silicon technolgy silicon technology which uses asilicon chip as a platen, has gained considerable
acce[tance since its commercial introduction in 1998.

Silicaon technology's strength
s include high image quality, approaching that of the better optical
devices: modest size requirementa allowing the technology to be integreated into small, low
devices, and potentially lower cost as a large number of silicon patens can be manufacture
d from a
singer wafer.

Ultrasound technologies

ultrasound devices are more capable of penetrating dirt and residue than optical and silicon devices

and are not subject to some of the image
dissolution problems found in larger optical devices.

Scan strengths

Proven technology capable of high levels of accuracy

Range of Deployment Environments

Ergonomic, easy
use devices

Ability to enroll multiple fingers

Scan weaknessess

Inability to enroll some users

performance deterioration over


association with forensic applications

Need to depploy specialized devices

Face recognition

• Ability for reliable face recognition is important to the security engineer because of the widespread
reliance placed on photo IDs. Applications for fa
ce matching include:

Identification and authentication;


Monitoring, fraud prevention.

• Systems rely on:

Face recognition performed by humans (most of the systems);

Automatic recognition.

Solutions for face recognition are based on

the following technologies:

• Neural networks;

• Eigenfaces

we'll take a closer look at this one;

• Local feature analysis.

“However, IBG's (International Biometrics Group) testing has found that the core technology is highly
susceptible to falsely nonm
atching users in one
one verifications and to failing to identify enrolled
users in one
many identifications.”


• Every face image can be treated as a vector in a highdimensional space of possible faces of human

• Suppose we have a
large set of images normalized to line up eyes and mouths, and resampled at the
same size in (say, m __n pixels.)

• The covariance matrix of the statistical distribution of face image vectors can be calculated.
Eigenvectors (of the size
) of this matrix

are called

• Eigenfaces with proper weights can be summed together to create an approximation for a given gray
scale image of a human face.

Eigenfaces may be improved significantly, if the background is removed.

The first task is to det
ermine whether the image is a face at all. This can be done by comparing it to a mean face
image and estimating thus “distance to the face space”.

• The system works as follows:

1)Define the face space: acquire a training set of images and calculate eigen
faces from them;

2)When a new face is encountered, project it onto the set of eigenfaces and calculate a set of weights.

3)Check if the image is a face at all by calculating the distance to the face space. Use a treshold to make

4) If it is a fac
e, classify the person as known or unknown by comparing the weight pattern of the image to the
known weight patterns. Simple Euclidean distance between weights vectors may be used.

5) (optional) Remember the face pattern, if it is new.
The system may be ma
de more “intelligent” by using
several images of the same person's face and by updating eigenfaces from time to time when new faces are

• 100 to 125 eigenfaces are usually used

Face recognition systems

• A facial recognition server controll
ing access at a facility with up to 30,000 persons would cost about $15,000.

• False Accept Rates vary from 0.3% to 5%, with False Reject Rates varying from 45% to 5%.

• The performance of systems may largely depend on:

camera distance


Expression of face

• Face recognition systems have 12% share of biometrics market.




Which can leverage existing camera technology.

It require specialized devices that provide necessary infrared illumination.

S/w components

Image processing