Biometric System

M.Daris Femila Dr.A.Anthony Irudhayaraj

Assistant Professor, Department Of Computer Science, Dean-Information Technology,
SRM Arts and Science College, Aarupadai Veedu Institute Of Technology,
Chennai, India Chennai, India
e-mail:darisbennison@gmail.com e-mail:anto_irud@hotmail.com

Abstract— This paper covers the field of biometric systems unique and immutable for each person and the basic
focus on biometric authentication systems. A short and general characteristics of fingerprints do not change with time. The
overview of biometric authentication systems gives some uniqueness of a fingerprint can be determined by the patterns
insight in how the various biometric data can be used for of ridges and furrows as well as the points on the surface of
authentication. The problems of varying biometric data, the finger. Minutiae points are local ridge characteristics that
caused by noise respectively human nature and approaches to occur at either a ridge bifurcation or a ridge ending.
solve these problems with multi-biometric systems in fingerprints are routinely used in forensic laboratories and
combination with information fusion, are also discussed. Since identification units all over the world and have been accepted
there is such a vast range of variances for the usage of
in the court of law for nearly a century. Since the 1980's the
biometric systems, some type of statistics is determined.
usage of fingerprints in civil areas has become more relevant
Keywords-biometric,verification,identification, authentication because of increasing accuracy and decreasing prices of
fingerprint devices. Some examples of the use of fingerprint
I. INTRODUCTION devices in civil areas are: Permitting logins based on
fingerprints.
First of all the term Biometrics should be more or less
defined in order to have a common understanding of the
subject. Both terms ”Biometrics” and ”Biometry” have been B. Handscan
used since early in the 20th century to refer to the field of This biometric approach uses the geometric form of the
development of statistical and mathematical methods hand for confirming an individual's identity. Specific
applicable to data analysis problems in the biological features of a hand must be combined to assure dynamic
sciences. Recently the term .Biometrics. has also been used verification, since human hands are not unique.
to refer to the emerging field of technology devoted to Characteristics such as finger curves, thickness and length,
identification of individuals on the basis of their biological the height and width of the back of the hand, the distances
traits, such as those based on retina-scans, iris-patterns, between joints and the overall bone structure, are usually
fingerprints or face recognition. The recent usage and extracted. Those characteristics are pretty much persistent
meaning of the term Biometrics will be the primary focus of and mostly do not change in a range of years.The verification
this paper. process requires the user to enter an ID in order to verify the
In today’s world a wide variety of applications requires claimed identity. After the user ID has been entered and the
reliable and secure authentication methods to confirm the photos have been captured, the calculation of the feature set
identity of an individual requesting their service. Some representing the biometric trait and the verification process
examples of such applications would include secure access lasts no longer than a second. Handscan applications have
to buildings, computer systems, laptops, cellular phones, proven their practical use which is shown by the 30-60%
memory such as USB sticks and many more. Furthermore it market share of biometric identification applications.
is possible to establish an identity based on who you are
rather than by what you posses. C. Signature
(e.g. identification cards) or what you remember.(e.g. Signature verification is the process used to recognize an
passwords). individual's hand-written signature. Dynamic signature
verification uses behavioral biometrics of a hand written
signature to confirm the identity of a person. This can be
II. BIOMETRIC SYSTEM EXAMPLES achieved by analyzing the shape, speed, stroke, pen pressure
and timing information during the act of signing. On the
A. Fingerprint other hand there is the simple signature comparison which
Among all the biometric techniques, fingerprint-based only takes into account what the signature looks like. So with
identification is the oldest method which has been dynamic signature verification, it is not the shape or look of
successfully used in numerous applications. The fingerprint the signature that is meaningful, it is the changes in speed,
itself consists of patterns found on the tip of the finger, thus
making it a physical biometric. Fingerprints are known to be
___________________________________
978-1-4244 -8679-3/11/$26.00 ©2011 IEEE

152
pressure and timing that occur during the act of signing, thus
making it virtually impossible to duplicate those features.
Devices which enable dynamic signature verification
store the behavioral factors and the captured signature image
itself for future comparison in their database. These devices
account changes in one's signature over time by recording
the time and the dynamic features each time a person uses
the system. The major difficulty with this technology is to
differentiate between the consistent parts of a signature,
these are the characteristics of the static image, and the
behavioral parts of a signature, which vary with each signing.
Comparing many signatures made by one individual reveals
the fact that an individuals signature is never entirely the
same and can vary substantially over an individual's lifetime.
Allowing these variations in the system, while providing the
best protection against forgery is a big problem faced by this
biometric technology.
The financial industry sometimes uses signature
verification for money transactions. The Manhattan Bank
was the first bank to test such an approach by using a
biometric signature application for their money transaction
system.

D. Iris
Iris scan biometrics employs the unique characteristics
and features of the human iris, which remains unchanged
throughout an individual's lifetime, in order to verify the
identity of an individual. The iris is the area of the eye where
the pigmented or colored circle, usually brown, green, grey
or blue, rings the dark pupil of the eye. The iris is well Fig. 1. Examples of some of the biometric traits associated with an
protected cause of the human anatomy and therefore injuries individual: (a) fingerprint, (b) face, (c) hand geometry, (d) signature, (e)
iris and (f) retina.
are rare. Typically the iris scan process begins with a
photograph which is taken with a special camera close to the
subject. The user has to be in between a maximum distance Research has proved that the patterns of blood vessels on
of about 1 meter to the reading device. The camera uses an the back of the human eye were unique
infrared imager to illuminate the eye and capture a very high
resolution photograph. E. Retina
The inner edge of the iris is located by an iris-scan Along with iris recognition technology, retina scan is
algorithm which maps the iris distinct patterns and perhaps the most accurate and reliable biometric technology.
characteristics. Systems using iris biometrics even work with It is also among the most difficult to use and requires well-
glasses and this technology is one of the few biometric trained, and is perceived as being moderately to highly
technologies that can work well in identification mode. Iris intrusive. The users have to be cooperative and patient to
patterns are extremely complex, carrying an astonishing achieve a proper performance. Basically the retina, a thin
amount of information and have over 200 unique spots. nerve on the back of the eye, is the part of the eye which
Unique spots are categorized into the tissue, which gives the senses light and transmits impulses through the optic nerve
appearance of dividing the iris in a radial fashion, rings, to the brain. Blood vessels used for biometric identification
furrows, freckles and the corona. The fact that an individual's are located along the neural retina which is the outermost of
right and left eyes are different and that patterns are easy to the retina's four cell layers.
capture, establishes iris-scan technology as one of the from person to person. It has even been proven that these
biometrics that is very resistant to false matching and fraud. patterns, even between identical twins, were indeed unique.
This pattern also doesn't change over the course of a lifetime.
Retinal scanners require the user to place their eye into some
sort of device and then ask the user to look at a particular
spot so that the retina can be clearly imaged. This technology
involves using a low-intensity infrared light source through
an optical coupler to scan the unique patterns of the retina.
The reflection of the vascular information is being recorded.
Retina scanning works well in both modes, identification and

153
verification. Additional advantages include the small In this mode biometric user data is acquired. This is
template size and good operational speed. mostly done with some type of biometric reader. Afterwards
the gathered information is stored in a database where it is
F. Voice labeled with an user identity(e.g. name, identification
Of the many types of biometric technologies available number) to facilitate authentication
today, voice identification and authentication solutions have
a unique edge over much of the competition, because B.Authentication mode
customers typically don't need to purchase new hardware to
implement the solutions. Most of the voice biometric Again biometric user data is acquired first and used by
solutions can be used through a typical telephone or the system to either verify the users claimed identity or to
microphone hooked up to the computer. In order to identify identify who the user is. While identification involves the
or authenticate users, most voice biometric solutions create a process of comparing the users biometric data against all
voice print of the user, a template of the person's unique users in the database, the process of verification compares
voice characteristics created when the user enrolls with the the biometric data against only those entries in the database
system. During enrollment the user has to select a passphrase which are corresponding to the users claimed identity. In
or repeat a sequence of numbers. The passphrase should be general one can consider the verification of the identity of a
in the length of 1 to 1.5 seconds. The problem with shorter person a two-class problem: either the person is who he/she
passphrases is that they have not enough data for claims to be (client) or the person fails to be the one he/she
identification. Longer pass phrases have too much claims to be (impostor) So we are basically dealing with a
information. The user has to repeat the passphrase or the binary-decision scheme where we either accept or reject a
sequence of numbers several time. person. Simple biometric systems usually consist of the
This makes the enrollment process lasting much longer following four components:
than with other biometric technologies. All subsequent
attempts to access the system require the user to speak, so 1) Sensor modules: This modules acquires biometric user
that their live voice sample may be compared against the data. Examples of sensor modules would be an retina
pre-recorded template. A voice biometric sample is a scanner or a fingerprint sensor.
numerical model of the sound, pattern and rhythm of an 2) Feature extraction modules: This modules is responsible
individual’s voice. A problem considering the voice is that for extracting feature values of a biometric trait. If hand
people's voices change over time along growth, or when geometry would be used as a biometric trait then feature
someone has got a cold or another disease. Background noise
values would include width of fingers at various locations,
can also be an disturbing factor.
width of the palm, thickness of the palm, length of fingers
etc.
G. Face
3) Matching modules: The matching modules compares the
Human face detection plays an important role in
applications such as video surveillance, human computer acquired biometric features against those stored in a
interfaces, face recognition, and face image databases . To database.
enable this biometric technology it requires to have at least a 4) Decision-making modules: The users identity is either
video camera, PC camera or a single-image camera. established or a claimed identity is accepted or rejected.
Nevertheless, this biometric approach still has to deal with a This is done based on the results of the matching modules.
lot of problems and cannot work with acceptable Since we are dealing with a binary decision scheme it is
identification rates unless certain restrictions are being obvious that the decision-making module can make two
considered. Finding a face in a picture where the position, kinds of errors. The errors, which can be made in the
the orientation, the background and the size of a face is process of verification, are called:
variable, is a very hard task and many algorithms have been False Rejection (FR): when an actual client gets identified
worked on to solve this problem. Other problems with face as an impostor
detection occur whenever faces are partially covered, as with False Acceptance (FA): when an actual impostor gets
beards, glasses, hair style or hats, because a lot of identified as a client.
information just stays hidden.
IV. PERFORMANCE EVALUATION
The performance of a biometric authentication system
III. BIOMETRIC AUTHENTICATION SYSTEMS
can be measured as the False Acceptance Rate FAR
Looking at biometric systems in a more general way will Equation (2), or the False Rejection Rate FRR Equation (1)
reveal certain things all biometric-based authentication which are defined as:
systems have in common. In general such systems work in FRR = number of false rejections/number of client
two modes: accesses ----(1)

A. Enrollment mode FAR =number of false acceptances/number of client

accesses ----(2)

154
 A perfect biometric authentication system would have a still may be impossible to acquire that trait because of the
FRR = 0 and a FAR = 0 which is a little bit unachievable in poor quality of the ridges which make up the fingerprint.
reality. It is also interesting that any of the two values FRR
and FAR can be reduced to an arbitrary small number, with VI. MULTI BIOMETRIC SYSTEMS
the drawback of increasing the other value. Another Most of the problems and limitations of biometrics are
interesting value is the TotalErrorRate TER Equation (3) imposed by unimodal biometric systems. Unimodal
which is defined as: biometric systems rely on the evidence of only a single
TER =number of FA + number of FR/total number of biometric trait. Some of these problems may be overcome by
access----(3) multi biometric systems and an efficient fusion scheme to
At this point it is important to emphasize the fact that combine the information presented in multiple biometric
these measures could be heavily biased by one or either type traits. It is evident that problems like non-universal traits,
of errors (FAR or FRR) depending only on the number of distinctiveness and security problems are easier and better to
accesses which have been used in obtaining these respective deal with if more biometric traits are present. So if a person’s
errors. This means that the TER will always be closer to that fingerprint cannot be acquired by a sensor, other biometric
type of error which has been obtained with the largest methods like voice recognition and retina scans are taken
number of accesses. The overall performance of a biometric into account and the resulting data is validated against the
authentication system should not be measured by the TER biometric database.
but rather by the Receiver Operation Characteristic ROC,
which represents the FAR as a function of the FRR. A. Fusion of Biometric data
In general there are three possible levels of fusion for
V. PROBLEMS WITH BIOMETRICS
combining two or more biometric systems to a multi
In theory collecting and verifying biometric data is no biometric system:
problem but in today’s demanding real-world applications 1)Fusion at the feature extraction level: Feature sets are
there are a lot of problems with biometric systems. One of acquired from each sensor where each feature set is
those problems is that biometric traits extracted from persons represented as a vector. Then the vectors are concatenated
tend to vary with time for one and the same person and to which results in a new feature vector with higher
make it even worse, this variation is itself very variable from dimensionality representing a person’s identity in a different
one person to another. Most of the other problems are caused hyperspace.
by extreme or constantly changing surroundings and the 2)Fusion at the matching score level: Each biometric
nature of certain biometric measures. system provides a matching score which indicates the
proximity of the feature vector with the template vector.
A. Noise Fusion at this level would mean combining the matching
Noisy biometric data like a person having a cold(voice scores in order to verify the claimed identity. In order to
recognition), a simple cut on ones finger(fingerprint scan) or combine the matching scores reported by the sensors,
different lighting conditions(face detection) are some techniques such as logistic regression is used. These
examples of noisy inputs. Other examples are misconfigured techniques attempt to minimize the FRR for a given FAR .
or improperly maintained sensors or inconvenient ambient 3)Fusion at the decision level: The resulting feature
conditions like dirt on a sensor for fingerprints or voice vectors from each sensor need to be classified into two
recognition with loud background noise. The problem with classes - reject or accept.
noisy biometric data is that authorized personnel may get Biometric terms, such as recognition, verification and
incorrectly rejected(FR), if the noisy data affects the identification, are sometimes used randomly. This is not only
extracted features so much, that no match can be found in the confusing, but incorrect as each term has a different meaning.
biometric database. The other extreme situation would occur Recognition is a generic term and does not necessarily
if noise would change the extracted features in such a way, imply either verification or identification. All biometric
that the result feature set would match to another person(FA). systems perform “recognition” to “again know” a person
who has been previously enrolled. Verification is a task
B. Distinctiveness where the biometric system attempts to confirm an
While a biometric trait is expected to vary significantly individual’s claimed identity by comparing a submitted
across individuals, there may be large similarities in the sample to one or more previously enrolled templates.
feature sets used to represent theses traits. Thus, every Identification is a task where the biometric system attempts
biometric trait has a theoretical upper bound in terms of to determine the identity of an individual. A biometric is
discrimination capability. collected and compared to all the templates in a database.
Identification is “closed-set” if the person is known to exist
C. Non-universality in the database. In “open-set” identification, sometimes
The problem of non-universality arises when it is not referred to as a “watchlist,” the person is not guaranteed to
possible to acquire certain biometric traits from all users. exist in the database. The system must determine if the
That means that even though a person has a fingerprint, it person is in the database. Because of these variances,
different statistics must be used for each task.

155
 VII.CONCLUSION
Biometric systems and especially multi biometric
systems have a huge potential of growth. By using biometric
technologies, access procedures should be made simpler,
faster and more secure. Especially governments, law
enforcement agencies, military and industrial companies,
already make partial use of this technology. In the future
biometric devices will surely become more involved in
many civil areas. Maybe in a couple of years access to ones
private home or car will be granted upon a successful iris
scan, thus making the traditional house or car keys obsolete.
Maybe money, credit cards and cheques will become
obsolete by leaving ones fingerprint instead of a certain
amount of monetary value. But in spite of all the advantages
coming along with the broader usage of biometric
technology in our everyday lives, this technology also
brings up a whole new range of difficulties and problems.
So it will not suffice to study factors like cost versus
performance tradeoffs, or usability and security issues
before deploying biometric systems. Very special care must
be taken what may be done with the acquired biometric data
and who may use it for a certain purpose.
REFERENCES
[1] A. K. Jain, A. Ross. .Mutlibiometric Systems., Communications of
the ACM, Vol. 74, pp. 34-40, 2004.
[2] A. K. Jain, A. Ross. .Information fusion in biometrics., Pattern
Recognition Letters, Vol. 24, pp. 2115-2125, 2003
[3] P. Verlinde, M. Acheroy. .A Contribution to Multi-Modal Identity
Verification Using Decision Fusion., Decision Fusion. ENST-Paris
Ph.D. Thesis, 1999.
[4] M. Roach, J.D. Brand, and J.S.D. Mason. .Acoustic and Facial
Features for Speaker Recognition., ICPR, 2000.
[5] A.K. Jain, S. Prabhakar, S. Chen. .Combining multiple matchers for a
high security fingerprint verification system., Pattern Recognition
Letters, Vol. 20, pp. 1371-1379, 1999
[6] S. Liu, M. Silverman. .A Practical Guide to Biometric Security
Technology.www.computer.org/itpro/homepage/Jan
Feb/security3.htm, 2000

156