|
|
|
|
|
|
| Biometric authentication is an
automated method whereby an individual's identity is confirmed by examining
a unique physiological trait or behavioral characteristic, such as a fingerprint,
iris, or signature. Physiological traits are stable physical characteristics,
such as palm prints and iris patterns. A behavioral characteristic-such
as one's signature, voice, or keystroke dynamics-is influenced by both controllable
actions and less controllable psychological factors. Although behavior-based
biometrics can be less expensive and less threatening to users, physiological
traits tend to offer greater accuracy and security. In any case, both techniques
provide a significantly higher level of identification than passwords or
smart cards alone. |
| The accuracy and performance of
biometric based system is measured by the following four criteria: FRR (False
Rejection Rate), FAR (False Acceptance Rate), FER (Failure to Enroll Rate),
and ERR (Equal Error Rate). Learn more about these values and their significance
in the selection
process section, where we discuss the different criteria determining
the selection of the 'perfect' authentication technology. |
| All biometric systems operate
in a similar fashion. First, the system captures a sample of the biometric
characteristic during the enrollment process. Unique features are then extracted
and converted by the system into a mathematical code. This sample is then
stored as the biometric template for the enrollee. |
| The following four-stage process
illustrates the way in which biometric systems generally operate:
|
- Capture - a physical or behavioral sample is captured during
enrollment.
- Extraction - unique data is extracted from the sample and a
template is created.
- Comparison/Match - the template is compared with a new sample
and the system determines whether there is a match.
- Storage - the system stores template data only on the server,
workstation, etc.
|
|
| At the heart of the biometric
system is the biometric engine, a proprietary element that extracts and
processes the biometric data. The system may apply an algorithm or an artificial
neural network. |
Return to
Top 
|
|
| Fingerprint recognition technology
is an authentication method in which captured fingerprint images are processed
through a series of image processing algorithms to obtain a clear unambiguous
skeletal image of the raw data, clarifying smudged areas, removing extraneous
artifacts and healing most scars, cuts and breaks. Unique physical characteristics
of the fingerprint, such as ridge ends and bifurcations (minutiae) within
the skeletal image are identified and encoded, providing critical placement,
orientation and linkage information for the matcher. |
| A typical fingerprint enrollment
and authentication process follows the the outlines below. A confirmed fingerprint
template never leaves a secured area within the system. |
|
Enrollment process:
|
Authentication process:
|
|
|
|
| The fingerprint's strength is
its acceptance, convenience and reliability. It takes little time and effort
for somebody using a fingerprint identification device to have his or her
fingerprint scanned. Studies have also found that using fingerprints as
an identification source is the least intrusive of all biometric techniques.
Several companies have produced capture units smaller than a deck of cards
or they have integrated the scanner into an existing devices such as a keyboard
or mouse. |
| One of the biggest fears of fingerprint
technology is the theft of fingerprints. Skeptics draw attention to the
fact that latent or residual prints left on the glass of a fingerprint scanner
may be copied. However, a good fingerprint identification device only detects
live fingers and will not acknowledge fingerprint copies. In order to avoid
many complete comparisons of fingerprints, a fingerprint is generally first
classified, or grouped, as belonging to a certain class of fingerprints. |
| Minutiae points are certain characteristic
points of fingerprint images, such as endings and bifurcations (splits)
of ridges. The mathematical information about the minutiae points is extracted
from the fingerprint image data by an appropriate feature extraction algorithm;
they are used as verification data and reference data for fingerprint recognition.
An individual is recognized as the lawful owner of the reference data if
they agree with the verification data with a predefined number of minutiae
points. The method of Vector Line Type Analysis is a newer method of extracting
and comparing fingerprint data. In contrast of the method of only considering
minutiae points, the whole fingerprint pattern is considered and a certain
number of points is classified by the present line types. |
Return to
Top
|
|
| Facial recognition systems can
automatically recognize persons by using a video camera to store and compare
facial features for authentication. The different face methods share certain
commonalities, such as emphasizing those sections of the face which are
less susceptible to alteration, including the upper outlines of the eye
sockets, the areas surrounding one's cheekbones, and the sides of the mouth.
All of the primary technologies are designed to be robust enough to conduct
1-to-many searches, that is, to locate a single face out of a database of
thousands, even hundreds of thousands of faces. |
| There are four primary methods
employed by facial scan vendors to identify and verify subjects include
eigenfaces, feature analysis, neural network, and automatic face processing.
"Eigenface," roughly translated as "one's own face," is a technology patented
at MIT which utilizes two dimensional, global grayscale images representing
distinctive characteristics of a facial image. Variations of eigenface are
frequently used as the basis of other face recognition methods.
|
| Feature analysis is perhaps the
most widely utilized facial recognition technology. One possible approach
is the use of Local Feature Analysis (LFA), which can be summarized as an
"irreducible set of building elements." LFA utilizes dozens of features
from different regions of the face, and also incorporates the relative location
of these features. The extracted (very small) features are building blocks,
and both the type of blocks and their arrangement are used to identify/verify.
It anticipates that the slight movement of a feature located near one's
mouth will be accompanied by relatively similar movement of adjacent features.
Since feature analysis is not a global representation of the face, it can
accommodate angles up to approximately 25° in the horizontal plane, and
approximately 15° in the vertical plane. Feature analysis is robust enough
to perform 1-1 or 1-many searches. |
Return to
Top
|
|
| Iris recognition refers to an
authentication method that utilizes the unique characteristics of the iris,
the colored portion of the eye that surrounds the pupil of the eye. The
iris contains many collagenous fibers, contraction furrows, coronas, crypts,
colors, freckles, rifts, and pits. Measuring the patterns of these features
and their spatial relationships to each other provides other quantifiable
parameters useful to the identification and verification process. The iris
recognition process begins with video-based image acquisition that locates
the eye and iris. The boundaries of the pupil and limbus are defined, eyelid
occlusion and specular reflection are discounted, and quality of image is
determined for processing. |
| An advantage of iris recognition
technology is that the user is not required to focus on a target because
the patterns of flecks are located on the surface of the eye. In fact, a
video image of the eye can be produced from up to three feet away, which
allows for the use of iris scanners at ATM machines, for example. Even visually
impaired persons with intact irises can generate images of their iris, which
can be captured and encoded with iris imaging products via active iris capture.
Since cataracts are a malady of the lens, which is behind the iris, they
do not affect the iris-scanning process in any way. Furthermore, eyeglasses
and contact lenses are easily accommodated. Iris recognition technology
vendors claim to offer the lowest FAR, highest accuracy, single-factor identification
method in the industry. |
Return to
Top
|
|
| Signature recognition is a behavioral
authentication method that uses an individual's signature as a means of
verifying the identity. Static signature capture is becoming a popular replacement
to pen and paper signing for bankcard, PC and delivery-service applications
(e.g., Federal Express). Signature verification is used with special devices
such as wired pens, pressure-sensitive tablets, or a combination of both.
Devices using only wired pens are less expensive and take up less room than
those employing pressure-sensitive tablets, but are potentially less durable.
To date, the financial community has been reluctant to adopt automated signature
verification methods for credit cards and check applications because signatures
are easily forged. This keeps signature verification from being integrated
into high-level security applications. |
| The biometrics of a handwritten
signature is based not only on the shape of the signature but more importantly
on selected signature dynamics. The signature is captured along with timing
elements (speed, acceleration) and sequential stroke patterns (did the "t"
get crossed from right to left and did the "i" get dotted at the very end).
These unique dynamics derived from a person's muscular dexterity are usually
referred to as "muscle memory." The brain, without any particular attention
to detail, automatically controls these nerve impulses. These "muscle memory"
dynamics yield accuracy results that are comparable to and less intrusive
than the alternative biometric technologies such as fingerprint.
|
Return to
Top
|
|
| Voice verification technology
verifies a person's identity on the basis of voice characteristics. These
unique features consist of cadence, pitch, tone, harmonics, and shape of
larynx. Biometric voice verification technology balances the need to tighten
security and reduce costs with the inherent convenience of the human voice.
Voice verification technology solves countless security problems that require
accurate user identity verification, by providing a powerful, non-intrusive
solution to protect both proprietary and confidential information. Many
biometric technologies are considered intrusive, and therefore, inconvenient
because they require the user to subject a part of their body to technical
scrutiny. Voice recognition has difficulties handling environmental factors
such as noise, and performance can be affected by colds or laryngitis, but
it offers a quick and convenient technology for integration into telecommunication-based
verification. |
Return to
Top
|
|
| Retinal recognition technology
is an authentication method that identifies individuals by sending a low-intensity
infrared light through the pupil to the blood vessel pattern on the "back"
of the eye. This beam of light is sent back to the scanning device, which
maps the capillary pattern of the retina, a thin nerve on the back of the
eye, into mathematical data, which is stored as a template and later used
to verify identity. With the current applications a user must keep their
head and eye motionless within a half-inch of the device, while focusing
on a small rotating point of green light. About 400 points of reference
are captured and stored, ensuring the measurement is accurate with a negligible
false rejection rate. |
| Unfortunately, the retinal scan
process is considerably more intrusive than other methods, such as an iris
scan, because it requires the user to remain quite close to the device and
it may reveal unneeded information (such as health issues). Hence, many
people are hesitant to use this technology. Also, a significant number of
people may be unable to perform a successful enrollment. In addition, there
are a number of degenerative diseases of the retina that can alter the scan
results over time and can negatively impact the retinal image quality. Considering
these limiting factors as well as a nearly 0% false acceptance rate (FAR),
retinal scanners are most often deployed in high-security access control.
Moreover, high system costs add to this technology's limited popularity.
|
Return to
Top
|
|
| Hand Geometry is based on the
fact that virtually every person's hand is shaped differently and that the
shape remains the essentially the same for life. This authentication technology
uses an optical scanner to record data. To use hand geometry, a user must
place their hand on a platen and positions it by lining it up with 5 guide
pegs. The typical image acquisition system comprises of a light source,
a camera, a single mirror and a flat surface (with five pegs on it). The
user places his hand - palm facing downwards - on the flat surface of the
device. The five pegs serve as control points for an appropriate placement
of the right hand of the user. Feature extraction involves computing the
widths and lengths of the fingers at various locations using the captured
image. The system examines up to 90 characteristics, including three-dimensional
shape of the hand, length and width of fingers and shape of knuckles.The
metrics define the feature vector of the user's hand. |
| A hand-geometry template is smaller
than those made by most other biometric authentication procedures, which
saves storage space and speeds up template retrieval time, but it also increases
the false acceptance rate (FAR). That is, given a large number of records,
this method may not be able to differentiate an individual from others with
similar hand characteristics. Hence, as the size of a database grows, there
must an increase in the number of distinguishing characteristics used for
authentication. Other disadvantages of hand geometry authentication include
a relative lack of hand uniqueness (as compared to other biometric characteristics,
such as fingerprints) and the possibility of hand deformation. Current research
work involves identifying new features that would result in better discriminablity
between two different hands, and designing a deformable model for the hand.
|
Return to
Top
|
|
| Keystroke dynamics, also known
as typing rhythms, refers to an authentication method that analyzes the
way a user types at a terminal by monitoring keyboard input 1,000 times
per second. This is a behavioral biometric authentication method in which
users enroll by typing the same word (or words), such as their usual user
name and password set, a number of times. The user must still remember the
user name and password. However, if anybody else discovers both of them
and tries to use them, the keystroke dynamics will stop the fraudulent user.
The key parameters are "flight time", the amount of time that a user spends
"reaching" for a certain key and "dwell time", the amount of time a user
spends pressing one key. The advantage in the computer environment is that
neither enrollment nor verification detracts from the regular workflow.
|
Return to
Top
|
|
