Journal of Theoretical and Applied Information Technology

30th June 2019. Vol.97. No 12

© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

COMPARATIVE AND ANALYSIS STUDY OF BIOMETRIC

SYSTEMS
FAOUZIA ENNAAMA1, KHALID BENHIDA1, AHMED BOULAHOUAL1

1
CadyAyyad University, High School of Technology, LAPSSII Laboratory, Safi-MOROCCO
E-mail: [email protected]

ABSTRACT

In recent years, there has been a growing interest in the field of biometrics as a powerful identification
technology. Various biometric technologies are based on behavioral and physiological analysis; therefore
they must be reliable, robust, simple and cheap.
In this paper, we have investigated an analytical comparison of different biometric systems namely:
fingerprint, iris, face, voice, keystroke dynamics, signature, retina, etc. and we have classified these
methods based on several criteria such as: universality, uniqueness, permanency, intrusiveness, effort, cost,
and reliability, as well as the most used biometric systems requested in the market and those that are of
greater interest in the current research work, and for each criteria we gave synthetic discussions.
Furthermore, we provided a brief overview of biometric methods, then we have described the modes used
in a biometric system such as enrollment, verification and identification and we have presented the possible
applications of biometrics.

Keywords: Biometric Systems, Identification, Recognition, Biometrics, Face.

1. INTRODUCTION biometrics is synonymous with physical

anthropology [10]. Another definition of biometrics
Today, determining the identity of an individual is given by Roethenbaugh [11] Biometrics are: “a
automatically is a problem that is still relevant. In a measurable characteristic or trait of a human being
daily environment that is becoming increasingly for automatically recognizing or verifying identity”.
interconnected, it is necessary to have automatic
In general, biometric characteristics have seven
and reliable authentication systems to identify users
factors that make them appropriate for biometric
and give them access to specific systems and
measurements and identification or authentication.
applications such as: e-commerce, banking, highly
These properties are [12]: (a) universality, which
secure places, etc.
means that every person should have the attribute,
Traditional systems based on the use of (b) uniqueness, theoretically refers that two people
passwords and identity cards cannot provide cannot have exactly the same characteristic, (c)
guaranteed authentication functions because they permanence, which indicates that the trait should
have a lot of weaknesses. In the first case, the be invariant with time, (d) collectability, which
password can be forgotten by the user or decoded indicates that the characteristic can be measured
by another person. In the second case, the badge (or quantitatively [12,13], (e) performance, which
identity card or key) may be lost or stolen. means the accuracy of the identification and the
Biometric systems are an alternative solution to the needed resources to achieve this accuracy must
two previous identification modes. respect the constraints imposed by the application,
(f) acceptability, which indicates, individuals who
Biometrics involves identifying a person from
use the application must present their biometric
one or more physiological characteristics
features to the system, and (g) circumvention,
(fingerprints [1], face [2], iris [3], hand geometry
which is about the easiest information to be
[4], retina [5], palmprint [6], etc.), or behavioral
collectible, measurable and usable for comparison.
characteristics (signature [7], gait [8], keystroke
dynamic [9], etc.). Etymologically, human

3466
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

Biometric systems are automatic systems

essentially using various modalities to identify or /
and recognize an individual [14], including
fingerprint, face, hand geometry, iris, retina,
signature, gait, palm print, voice model [15], ear
[16], vein of the hand [17], DNA [12], etc.
In fact, a practical biometric system must have an
acceptable accuracy and a reasonable recognition
speed with respect the required resources, harmless
for the users, accepted by the population, and robust Figure 1: Block diagram of a biometric system in
enough against fraudulent methods [18]. enrollment mode.
This paper provides a comparative and analytical  Verification: This is a one-to-one comparison
study of biometric systems. This investigation step (usually 1:1) [19], it consists of four main
based on the review of the biometric methods, their methods namely: data acquisition, pre-
recent developments and their statistical analysis of processing, feature extraction and matcher.
submitted publications rate in a scientific journal This mode verifies the identity of an individual
will be discussed in detail. We have chosen to by comparing the extracted biometric signature
organize this paper around five parts: The first with his or her own pre-recorded biometric
section is dedicated to Biometric system and its template in the system database. If it is correct,
possible modes commonly used. The second section he accepts it otherwise he rejects it.
gives a description of biometric methods; the third
section is about the applications of the most
requested biometric systems on the market. The
comparison results will be discussedin the fourth
section and the conclusion will be given in the last
section.
2. BIOMETRIC SYSTEM AND ITS MODES
A biometric system is an automatic system based
on the recognition of the biometric characteristics Figure 2: Block diagram of a verification system.
of an individual.
 Identification: consists of comparing the
Generally, a biometric system can operate in
characteristic in question, often called the
three modes [19]: in enrollment mode, in user's model with the equivalent models of all
verification mode or in identification mode. the users, already stored in a database. It is a
 Enrollment: This is the first step of a biometric comparison "1 to N" [19]. The identification is
system. It aims to collect biometric information also composed of four sub-parts such as: data
of the individual to identify him. Several data acquisition, pre-processing, feature extraction
acquisition steps must be performed in this and matching (One to Many). These subparts
mode namely; data acquisition, pre-processing work in the same way as in the case of
and feature extraction. The data acquisition is verification, except in a matching phase. In this
done by a biometric sensor to collect the mode, the comparison is done as one by one
biometric data of the individual. Then, in the while in the verification, all the data stored in
pre-processing stage, these characteristics will the database will be compared to have a better
be represented in digital form (signatures), possible correspondence with the model of the
after the suppression of the different sources of user. This mode involves associating an
noise of the individuals collected data. identity with a person.
Furthermore, this information is stored in a
specific database. We note that the processing
related to enrollment is not subject to time
constraints, since it is done in "offline" mode.

Figure 3: Block diagram of an identification system.

3467
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

3. BIOMETRIC METHODS forensic applications into civilian applications

[22]. With strong recognition performance and
A biometric system automatically identifies an
the growing market for low-cost personal
individual based on two main biometric
computers and acquisition devices, fingerprint
modalities: physiological and behavioral.
biometric systems are becoming very popular
 Physiological Biometrics: This type is based on and may be used in a variety of applications
the identification of particular physical traits such as PC login, e-commerce, ATMs, physical
that for any individual are unique and access control, etc.The principle of each
permanent. This category includes: fingerprint, biometric fingerprint system consists first of all
contour of the hand, face, retina, DNA, iris, in capturing and recording the image of the
etc. finger by an acquisition system. This image is
 Behavioral biometrics: This type is based on then transmitted to a processing system that
the analysis of certain behaviors of an will analyze it and compare it with those
individual such as the tracing of his signature, already recorded. The data acquisition is done
his gait and his keystroke dynamic. by an electronic sensor of optical, thermal,
In this part, we present a brief introduction to capacitive or ultrasonic type [1].Subsequently,
some commonly used biometric modalities; figure the captured image of the fingerprint will be
4 shows some example of biometric modalities. initially binarized to increase the contrast,
reduce the volume and increase the speed of
processing. Then the system extracts the
minutiae. This extraction makes it possible to
establish a template of the imprint that will be
compared to that already stored. The
identification or recognition is performed by
comparing the template of the test fingerprint
with all previously saved templates.
 Iris: The iris is the colored part of the eye; it
Figure 4: Some examples of biometric modalities [20]. should not be confused with the retina [23], the
back of the eye, which is also used as
 Fingerprint: is the design represented by the biometrics. It is an extremely reliable technique
ridges and valleys of the epidermis. It consists because it contains an infinite number of
of a set of locally parallel lines forming a characteristic points [24].So we could call it an
unique and different pattern for each individual optical fingerprint [25]. For all these reasons,
[1]. We differentiate the streaks or ridges - the the iris makes it possible to distinguish
lines in contact with a surface to the touch - different people, even in the case of
and the valleys - the hollows between two monozygotic twins.In addition, the texture of
streaks -. The streaks contain in their center a the iris of the right and left eyes belonging to
set of regularly spaced pores. Each imprint has the same person are different [25]. Most of the
a set of global (centers and deltas) and local work on iris identification and verification was
(minutiae) singular points [21]. The first traces done in the 1990s [26]. However, early systems
of human fingerprinting have been discovered have limited capacity to recognize the identity
on a large number of archaeological artifacts of the individual with precision and efficiency,
and historical items [1]. Although these so that much more needs to be done to improve
discoveries prove that ancient peoples were some technologies by adding other
aware of the individuality of fingerprints, this performances from a practical point of
scientific technique was not approved and view.The main difficulty of human iris
initiated until the end of the sixteenth recognition is that it is difficult to find apparent
century.At the beginning of the 20th century, characteristic points in the image of the first
fingerprint recognition was officially accepted sight.The acquisition of the iris is done by
as a valid personal identification method and means of a camera to compensate for the
became routine in forensic science [22]. inevitable movements of the pupil. It is very
Worldwide fingerprinting agencies and sensitive (precision, reflection ...) and
criminal fingerprint databases have been relatively unpleasant for the user because the
created. Automatic fingerprint recognition eye must remain wide open and it is
technology has evolved rapidly beyond

3468
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

illuminated by a light source to ensure proper diatopic (depending on the region) and
contrast. diaphasic (stylistic or situational) variations
 Face: This method looks at the face shape. [29].Voice capture is relatively easy to perform
Robust systems of facial recognition are in using a microphone, but is sensitive to ambient
great demand in several applications against noise [30]. It should be noted that this
crimes and terrorism, as well as the access technique is nevertheless difficult to use
control of personnel regardless of physical or because it is extremely subject to external
virtual space.Moreover, face recognition is a conditions (illness, stress of the person, etc.)
non-intrusive method, and facial attributes are [28, 29, 30]. Because of this, it is both a
the most common biometric characteristics physical and a behavioral characteristic. It is
used by human beings to recognize each sometimes chosen in combination with another
other.The most popular approaches to face characteristic (voice and writing for
recognition [2] are based on: (i) the location example).On the other hand, in everyday life,
and shape of facial attributes, such as the eyes, identifying a person through a recording of his
eyebrows, nose, lips and chin, and their speech is an ordinary task that can be found in
relationships and spatial distances; (ii) the many cases. For example, a listener may know
overall analysis of the facial image: which a speaker's voice easily either over the phone
represents the image of the face as a linear or in a crowd of individuals. Also, in court
combination of its characteristics.Although the applications, a voice recording may be the only
authentication performance of commercially evidence to recognize the identity of the
available facial recognition systems is criminal.The main purpose of voice
reasonable [27], it imposes a number of recognition is to extract the characteristics
limitations on how facial images are obtained, contained in a speech signal that is physically
often requiring a fixed and simple background like a variation of the air pressure produced by
to controlled lighting.These systems also have the articulator system [31].This signal will be
difficulty matching face images captured from converted by sampling into an electrical signal
two different views, under different lighting obtained at the output of a microphone, so that
conditions and at different times.For a face it can be compared to models in digital form.
recognition system to function properly in Then it will be quantified in discrete values of
practice, it must automatically (i) detect amplitude.Then, the noise of the acoustic
whether a face is present in the acquired image; signal must be removed and made very clear in
(ii) locate the face if there is one; and (iii) order to be transformed into a series of vectors
recognize the face from a general point of view typically having from 8 to 50 coefficients [31].
(i.e. any pose) under different environmental Subsequently, it is generally necessary to
conditions. reduce the size of the acoustic vector by
 Voice: The voice is an interesting biometric keeping the maximum of the characteristic
feature and is extremely sought after in systems information of the original signal.
related to remote applications where the  Hand Contour (Hand geometry): This is a
individual authenticates himself by telephone popular technology that is widely used for
for example. Voice biometrics aims to identify physical access control or clocking[32].
the idiosyncratic characteristics of the speech Historically, Jain and Duta [33] were the first
signal, produced at a given moment and under to analyze a deformable shape and develop a
specific conditions.Indeed, the production of method in which hand shapes are stored and
speech is a complex process that is influenced compared according to the average
by several factors at different levels [28]. As a misalignment.The main advantage of this
single speech sample represents several types modality is that it is simple and easy to capture.
of information, both linguistic and typical In addition, the biometric system is not very
features encapsulated together.The attributes of sensitive to the state of the hand, in other
the speaker certainly include clues related to words, dirt and cuts will not prevent its
the physical properties of his vocal organs. operation [32].However, the geometry of the
Furthermore, many distinctive non- hand also has disadvantages that can make the
physiological features of the speaker mark the system imprecise. Indeed, different individuals
acoustic signal, such as information about the can have almost similar hand shapes. In
emotional state of the speaker. In addition, addition, jewelry and accessories can be a
speech is influenced by diastatic (enzymatic), challenge to extract shape information from the

3469
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

hand.Also known as hand measurements in biometric modality such as the iris.The

literature, hand geometry has been the subject acquisition of the retina is as follows: The
of much attention from hand biometrics. As its person must place his eye in front of a capture
name indicates, the user positions his hand on hole located on the acquisition device [40]. A
the system to take different geometric light beam passes through the eye to the
measurements, namely the size of the palm, the capillary blood vessels of the retina. The
length of the hand or the fingers, the width and system locates and captures about 400
the perimeter of the hand and the fingers, etc.In reference points [40]. This technique requires
particular, Luque-Baena et al. [34] extracted close cooperation on the part of the user, as he
403 geometric features, including areas, must place his eye extremely close to the
perimeters, circularity measurements, camera. This technology is very secure and
compactness, etc. This is actually the highest reliable, which is why it is used in high-end
number adopted in the literature to improve the security applications such as military
performance of the personalrecognition system. installations and power plants.
Although geometric features are easy to  Ear:is a new viable biometric class because it
extract, they are not discriminating enough to has desirable properties such as universality,
be used for high security identification or uniqueness and permanence [41]. For example,
verification situations.In fact, the error rate in the ear is rich in features; it is a stable structure
hand recognition is quite high, especially for that does not change with age; thus, it does not
people belonging to the same family because of change shape with facial expressions,
a strong similarity. Moreover, the shape of the cosmetics and hairstyles. The ear is largely
hand changes a lot with age [35].Therefore, compared with the iris, retina, and fingerprint
some authors suggest the fusion of geometric and therefore is more easily captured at a
features with other features of the hand such as distance [14].Although it has advantages over
palm-prints [36] or finger shapes [37]. other biometric systems, it has not attracted
 Palmprint: is one of the most commonly used attention compared to other biometric systems
biometric recognition techniques used for such as face, fingerprint and gait.
crime.The palm print represents the inner part  Gait:indicates the way a person walks. This
of the hand (the invisible part when the hand is method is one of the few biometric features to
closed) between the wrist and the base of the recognize people at a distance [8].Therefore,
fingers. The use of the characteristics of the this attribute is very appropriate in surveillance
palmprint in the identification process was scenarios where the identity of an individual
originally proposed by Shu and Zhang [38]. can be established confidentially. Most gait
The modality of the palmprint impression recognition algorithms extract the human
contains different types of characteristics that silhouette in order to remove the spatio-
can be exploited for the recognition of the temporal characteristics of a moving
person [38]: (1) Determine vertical line individual. It is for this reason that the choice
segments, (2) Detect the secondary wrinkles or of a good model to represent the human body is
folds and (3) Determine the ridges.All these essential to the effective functioning of a gait
properties can characterize a person because recognition system.Gait based systems also
they are stable over time. Regarding wrinkles provide the ability to follow a person for a long
or secondary lines, they are thinner and more time. However, an individual's gait is
irregular than the main lines. High resolution influenced by several factors, including the
imaging is required for good extraction of choice of shoe, the type of clothing, leg
minutiae. The retinal vascular system is rich in support, walking surface, etc [42].
structures. It is claimed to be the safest  Signature: is the way in which a person signs
biometric modality such as the iris. his name to be a characteristic identifying that
 Retina: This is an old technique that uses the individual [7]. This handwritten is a biometric
iris. It is based on the fact that the blood behavioral technique that can change over
vessels of a retina are unique for each person. time, influenced by the physical and emotional
Even, the structures of the right and left eyes conditions of the signatories [43]. This type of
are different.The retina of a person is unique biometrics is currently little used but its
and remains unchanged throughout life defenders hope to impose it quickly enough for
[39].The retinal vascular system is rich in specific applications.Although signatures
structures. It is claimed to be the safest require user contact with a writing tool and a

3470
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

graphics palette, they have been accepted in b. Non universality: although biometric
government, legal, and commercial modalities exist in every individual, there
transactions as an authentication method.The are exceptions in which a person is not
signature processcontains a graphic palette able to present his biometric trait due to
with a pen. This device will measure several pathological conditions or a working
characteristics during the signature, such as environment characterized even by
speed, order of strikes, pressure and erasing fingerprint data or palmprint. For
accelerations, total time, etc [7]. The major example, a fingerprint system may fail to
disadvantage of handwritten signature is that capture the friction peak structure of some
even if the signatures of the same person can individuals due to the poor quality of their
change dramatically, thieves can produce the fingerprints. Similarly, an iris recognition
same signatures that fool the signature system may be unable to obtain
verification system [14]. information about a person's iris because
 Keystroke dynamics: This technology doesn’t of their long eyelashes or certain
require any special equipment since every pathological conditions of the eye.
computer has a keyboard. It is not unique to c. Upper bound on identification accuracy:
each individual but can be expected to provide (also called Temporal variability and non
sufficient discriminatory information to enable uniqueness) that is when the system
identity verification.It is a software device that becomes limited by the observed
calculates the amount of time a finger presses a variations in the set of characteristics of
key and the time a finger is in the air (between each subject (ie, intra-class variations)
keystrokes) [9]. This measurement is captured and by the variations between sets of
about a thousand times per second. The typing characteristics of different subjects (that
sequence is predetermined in the form of a is, inter-class variations). For example, in
password. Initially the person must compose the identical twin identity case, the main
his password so that a template is created. This lines of their palms can lead to an
biometric device is used as a verification inaccurate match due to incorrect data
method for electronic commerce and as a inducing a false rejection.
mechanism for controlling access to databases d. Spoof attacks:biometric traits like voice,
[12].We can say that Keystroke dynamics fingerprint and signature are easily
permits “continuous verification” of an vulnerable to spoofing attacks. For
individual's identity over a session after the example, the voice can be imitated and
individual logs in using a performing biometric fingerprints can be falsified for identity
like fingerprint or iris [14]. theft.
 Multimodality: Multimodality is defined as In this regard, some of the first multimodal
the use of different biometric systems. The biometric systems reported in the literature
most importantmain to combine various combined the face and voice characteristics of
biometric systems is to reduce the limitations individuals [44]. There are other works that
of single biometric systems. In fact, the combine several modalities such as: ear and
combination of different biometric systems face in [45], hand geometry and palmprint in
aims to ameliorate recognition performances [46], etc.
by increasing the quantity of discriminant data
of each individual, and to have a very low 4. APPLICATIONS OF BIOMETRIC
failure.However, biometric systems have so far SYSTEMS
failed to achieve accurate recognition. Several Nowadays biometric systems are increasingly
challenges confronted biometric systems are integrated in several applications. These
presented below [14]: applications can be classified into three main
a. Noise effect: The biometric information groups [47, 48]:
presented to the system can be infected by • Business applications: such as computer network
noise due to imperfect acquisition opening, electronic data security, e-commerce,
conditions or subtle variations in the Internet access, credit card, physical access control,
biometric value itself. These variations cell phone, registry management medical, distance
can cause bad illumination as the case of learning, etc.
the iris and the face or the dirty surface of
the sensor for the case of the fingerprint.

3471
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

• Government applications: such as national For the performance, the number of stars is related
identity card, driver's license, social security, border to the value of the Equal Error Rate (EER) obtained
control, passport control, etc. in the state of the art [13].
• Legal applications: such as body identification, 5.1.1 Discussion 1
criminal investigation, terrorist identification, etc. According to this table, we note that no modality is
perfect and therefore is not supposed to respond
5. COMPARISON RESULTS OF VARIOUS effectively to all the requirements (for example,
BIOMETRIC SYSTEMS permanence, acceptability, community) imposed by
all applications (Digital Rights Management
Each physiological or behavioral biometric
(DRM), access control, social protection
modality has its strengths and weaknesses, and the
distribution). For example, DNA is one of the most
choice depends on the targeted application. In other
effective techniques for verifying the identity of an
words, no biometric system is "optimal" and to be
individual. But, it cannot be used for access control
effective, it must have the ability to adapt to the
for reasons of computation time and also because
permanent and temporary changes of the user [14,
the individual would not be ready to take a little
47].
blood to carry out the identification. The choice of
5.1 Comparison according to biometric the modality is therefore dependent on the
modalities properties application used.
In this part, we will give a comparative analysis we can also point out that the iris, fingerprint,
concerning the biometric modalities based on retina, hand veins are very reliable thanks to their
several studies. We start with Jain et al. who made uniqueness and performance; but presently, they are
a comparison according to properties mentioned not frequently used because they are thought to
above (see Table 1). represent an intrusion into one’s private life and
Table 1: Comparison of the biometric modalities their fabrication costs are very expensive
according to the following properties: (a) Universality, specifically for the iris, retina and hand veins.
(b) Uniqueness, (c) Permanency, (d) Collectability, (e) Methods such as the face, voice, contour of the
Acceptability, (f) Performance [13]. hand are very acceptable by individuals by reason
Biometric a b c d e f of their ease of use, and their biometric systems are
modality not expensive to manufacture.In the following part,
DNA Yes Yes Yes Not Not really **** we will give an analytic comparison according to
really *
Blood Yes No Yes Not No * others performances.
really
Brain signal Yes Yes Yes Not No **** 5.2 Analytic comparison according evaluation
really criteria
Heart signal Yes Yes Yes Not Yes ****
really * Several studies have been conducted to evaluate the
Signature Yes Yes Not Yes Yes **** performance of biometric systems. The American
really company - the International Biometric Group [IBG]
Gait Yes No Not Yes Yes ***
[49] - for example has carried out a study based on
really
Keystroke Yes Yes Not Yes Yes **** four evaluation criteria:
really  Intrusiveness: the existence of a direct contact
Haptic Yes Yes Not Not Yes **** between the sensor used and the individual to
behavior Really Really be recognized.
Voice Yes Yes Not Yes Yes ****
Really  Reliability: effectiveness of the method (ability
Iris Yes Yes Yes Yes Somewhat **** to identify someone).
*  Cost: cost of the technology (readers, sensors,
Retina Yes Yes Yes Yes Somewhat **** etc ...), must be moderate.
*
Face Yes No Not Yes Yes ****  Effort: required by the user when entering
really biometric measurements. That is, techniques
Hand Yes No Yes Yes Yes **** that don’t need enough effort by the user.
geometry The comparison results of the most used biometric
Hand veins Yes Yes Yes Yes Yes ****
* methods used namely: contour of the hand,
Ear Yes Yes Yes Yes Yes **** fingerprint, face, voice, retina, iris, signature and
* Keyboard Strike are presented as follows [46]:
Fingerprint Yes Yes Yes Yes Yes ****  The least "intrusive" techniques to the most
"intrusive": voice, Keyboard Strike, signature,

3472
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

contour of the hand, face, fingerprint, iris and disadvantage of being intrusive, which greatly
finally retina. limits their areas of application. In addition, a
 The most reliable techniques to the least method such as the identification of the iris remains
reliable: iris, retina, fingerprint, face, hand, binding for the users. Conversely, identification
voice, and finally at an equivalent level, systems based on the analysis of face images pose
Keyboard Strike and signature. no difficulty for users. Face recognition is a method
 The least expensive to the most expensive that can be implemented independently of other
techniques: Keyboard Strike, voice, signature, biometric modalities, and is often used in
fingerprint, face, contour of the hand, retina surveillance applications.
and finally iris. 5.3 Statistical analysis of publications on
 The easiest techniques of use to the most biometrics
difficult. Here, it is worth appreciating the Another comparative study was carried out by
degree of possible interaction with the system: Prabhakar et al. [50] about statistical analysis of
face, signature, iris, Keyboard Strike, voice, publications specific to biometric techniques
fingerprint, contour of the hand and finally submitted and published in the scientific journal
retina. IEEE Transaction on PAMI. This analysis
On the same subject, the International Biometric compares face, fingerprint, Iris, multimodal
Group has carried out a comparative study on the biometrics and the “other” category includes: ear,
most popular and most used methods in the security signature, EEG (Electroencephalography), gait,
market. The results are presented in the form of a hand, dental, and speaker. Besides, it presents for
graph as shown in the figure 5. each subject the percentage of Paper
Distributionrelated to one of the above subjects.
Table 2 shows the distribution results of submitted
and accepted papers[50].

Table 2: Percentage Distribution of Submitted and

Accepted Papers by Biometric Subarea [50].

Paper distribution [%] Submitted Accepted

Face 33% 32%
Fingerprint 17% 16%
Multimodal 16% 16%
Iris 9% 11%
Figure5:Comparative Biometrics Market Share: 2001 Performance 4% 5%
[49]. Evaluation
5.2.1 Discussion 2
From figure 5, we can note that the most Other 21% 20%
widespread technologies are fingerprints (48.8% of
the market) followed by face recognition (15.4% of This study shows that the face recognition is the
the market), hand geometry (10.4%), iris (6.2%), most published technique at the level of the
voice recognition (4.3%), retina; ear morphology scientific journal IEEE Transaction with a
and Dental radiography represent the remainder of percentage of 33% of the total number of
the percentage. It should be noted that behavioral publications, followed by the fingerprint (17%),
techniques (such as gait, smile, eye movement, etc.) multimodal systems (16%), Iris (9%) and finally
are hard for them to overcome. other category with (22%) because there are four
Note that fingerprints are the most used biometric papers on the topics of: handwriting, ear, and two
characteristics. The first automatic authentication on brain waves or EEG [50].
system using fingerprints was marketed in the early
1960s. In addition, several studies have shown that 5.3.1 Discussion 3
the iris is the most reliable feature because the Face recognition made up the majority of
structure of the iris remains stable over the course submissions and accepted papers. There are several
of life. However, these methods have the major factors which make face recognition one of the best
biometric methods. It is a task that humans do

3473
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

habitually and naturally in their daily lives, non- This paper presents an analytic comparison of the
intrusive, socially well accepted and easily main biometric technologies. This comparison is
implemented unlike most biometric modalities, based on several properties like: universality,
which necessitate professional materiel during their uniqueness, permanency, collectability,
implementation (sensors and scanners for acceptability and performance or their evaluation
fingerprints, palm prints, retina and iris), face criteria such as: intrusiveness, effort, cost, and
image can be easily captured via digital cameras. reliability, as well as the most used biometric
Currently, 3D facial representation seems to be a systems requested in the market and those that are
promising method to deal with many of the human of greater interest in the current research work.
face variations such as illumination [51], pose Fingerprint biometrics remains the most widely
changes [52] and varying facial cosmetics [53]. used technology around the world and is the most
Moreover, it allows recognizing the individuals at a requested in the global market. It is a reliable and
distance thanks to surveillance cameras [54] and intrusive technique. The iris technique is one of the
identifying the face of the subject from a technologies (with the retina) that ensure a high
multifaceted image [55] contrary to other methods level of security. The iris provides a very high
(need contact with the sensors). For these reasons uniqueness and extended stability, resulting in
and others the scientific researchers are interested extraordinary reliability. These three methods
in the face recognition. Fingerprint and iris require a great cooperation by the users because of
recognition methods are extremely reliable their intrusiveness which makes them unacceptable
technique. However, they demand much explicit to users compared to the other methods.
effort from users. Especially, fingerprint requires Biometry by the contour of the hand is simple to
that the user puts down his finger to have well implement. However, it is not a very reliable
physical contact with the surface of the sensor to technique. The ear biometric is relatively new and
acquire the image, while iris acquisition currently not yet widely studied and has not attracted very
requires that the individual cooperate to careful attention by the scientific community.
position their eye relative to the sensor. The first Face recognition is a non-intrusive, natural, easy for
problem that now arises with these two techniques implementation and less costly method. It is also
is to leave the surface of the sensor clean because requested in the market and well accepted by users.
by dint of use it becomes dirty. The second problem Besides it currently acquires a greater interest to the
is that the two techniques invade one's private life scientific researchers. All of these factors make face
which makes it inacceptable. There is therefore a recognition one of the best biometric methods.
very strong demand for improved performance in Finally, behavioral methods such as signing and
face recognition systems and increasingly important typing are still known methods but still don’t have
in the field of research. very high identification accuracy.
Although it has desirable properties such as
universality, uniqueness and permanence, the ear REFERENCES:
biometric is relatively new and not yet widely
studied and has not attracted attention compared to [1] Maltoni, Davide, Maio, Dario, Jain, Anil K., et
other biometric systems such as face, fingerprint al. Handbook of fingerprint recognition.
and gait. Springer Science & Business Media, 2009.
For the signature, it has the advantage over other [2] Jain, Anil K. and Li, Stan Z. Handbook of face
biometric measures of being commonly used for recognition. New York : springer, 2011.
transactions. For this reason, the signature as a [3] Daugman, John. How iris recognition works.
means of identification is generally well accepted. In : The essential guide to image processing.
The problem of signature recognition comes from Academic Press, 2009, pp. 715-739.
the great variability that exists between two [4] Kumar, Ajay And Zhang, David. Hand-
occurrences of the signature of the same individual. geometry recognition using entropy-based
In addition, the signature may be affected by the discretization. IEEE Transactions on
health or emotional state of the individual or can be information forensics and security, Vol. 2, No
falsified by other person. 2, 2007, pp. 181-187.
6. CONCLUSION [5] Sukumaran, S. And Punithavalli, M. Retina
recognition based on fractal
Biometric recognition is a powerful tool for dimension. IJCSNS Int J Comput Sci and Netw
increasing security in the private and public sectors. Secur, Vol. 9, No 10, 2009, pp. 66-7.

3474
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

[6] Shu, Wei and Zhang, David. Palmprint evaluation. Springer Science & Business
verification: an implementation of biometric Media, 2005.
technology. In : Proceedings. Fourteenth [19] Jain, Anil K., Ross, Arun, Prabhakar, Salil, et
International Conference on Pattern al. An introduction to biometric
Recognition (Cat. No. 98EX170). IEEE, 1998, recognition. IEEE Transactions on circuits and
pp. 219-221. systems for video technology, Vol. 14, No 1,
[7] Lee, Luan L., Berger, Toby, and Aviczer, Erez. 2004.
Reliable on-line human signature verification [20] Idrus, Syed Zulkarnain Syed, Cherrier, Estelle,
systems. IEEE Transactions on Pattern Rosenberger, Christophe, et al. Soft biometrics
Analysis & Machine Intelligence, No 6, 1996, for keystroke dynamics. In : International
pp. 643-647. Conference Image Analysis and Recognition.
[8] Lee, Lily, and Grimson, W. Eric L. Gait Springer, Berlin, Heidelberg, 2013. pp. 11-18.
analysis for recognition and classification. In [21] Babler, W. Embryologic development of
: Proceedings of Fifth IEEE International epidermal ridges and their configurations. Birth
Conference on Automatic Face Gesture defects original article series, Vol. 27, No 2,
Recognition. IEEE, 2002, pp. 155-162. 1991, pp. 95-112.
[9] Monrose, Fabian and Rubin, Aviel D. [22] Gaensslen, Robert E., Ramotowski, Robert,
Keystroke dynamics as a biometric for and Lee, Henry C. Advances in fingerprint
authentication. Future Generation computer technology. CRC press, 2001.
systems, Vol. 16, No 4, 2000, pp. 351-359.
[23] Ross, Arun. Iris recognition: The path
[10] National Research Council, Whither forward. Computer, Vol. 43, No 2, 2010, pp.
Biometrics Committee, et al. Biometric 30-35.
recognition: challenges and opportunities. [24] Lim, Shinyoung, Lee, Kwanyong, Byeon,
National Academies Press, 2010. Okhwan, et al.Efficient iris recognition through
[11] Roethenbaugh, Gary. An Introduction to improvement of feature vector and
Biometrics and General History. Biometrics classifier. ETRI journal, Vol. 23, No 2, 2001,
Explained, Section, Vol. 1, 1998. pp. 61-70.
[12] Jain, Anil K., Bolle, Ruud, and Pankanti, [25] Davson, Hugh. Physiology of the Eye.
Sharath (ed.). Biometrics: personal Macmillan International Higher Education,
identification in networked society. Springer 1990.
Science & Business Media, 2006. [26] Daugman, John G. Fast focus assessment
[13] Wayman, James, Jain, Anil, Maltoni, system and method for imaging. U.S. Patent No
Davide, et al. An introduction to biometric 6,753,919, 22 juin 2004.
authentication systems. In : Biometric Systems. [27] Phillips, P. Jonathon, Grother, Patrick,
Springer, London, 2005. pp. 1-20. Micheals, Ross J., et al. FRVT 2002: Overview
[14] Flynn, Patrick J., Jain, Anil K., and Ross, Arun and summary, 2003.
A. (ed.). Handbook of biometrics. Springer, [28] Rabiner, Lawrence R. and Schafer, Ronald
2008.
W. Digital processing of speech signals.
[15] Campbell, Joseph P. Speaker recognition: A Englewood Cliffs, NJ : Prentice-hall, 1978.
tutorial. Proceedings of the IEEE, Vol. 85, No
[29] Labov, William. Sociolinguistic patterns.
9, 1997, pp. 1437-1462. University of Pennsylvania Press, 1972.
[16] Chen, Hui and Bhanu, Bir. Human ear
[30] Scherer, Klaus R. Vocal Affect Expression: A
recognition in 3D. IEEE Transactions on review and a model for future
Pattern Analysis and Machine Intelligence, research. Psychological bulletin, Vol. 99, No 2,
Vol. 29, No 4, 2007, pp. 718-737.
1986, pp. 143.
[17] Ding, Yuhang, Zhuang, Dayan, and Wang, [31] Boite, René. Traitement de la parole. PPUR
Kejun. A study of hand vein recognition
presses polytechniques, 2000.
method. In : IEEE International Conference
Mechatronics and Automation, 2005. IEEE, [32] Ross, Arun, Jain, Anil, and Pankati, S. A
2005, pp. 2106-2110. prototype hand geometry-based verification
system. In : Proceedings of 2nd conference on
[18] Wayman, James L., Jain, Anil K., Maltoni, audio and video based biometric person
Davide, et al.(ed.). Biometric systems:
authentication. 1999, pp. 166-171.
Technology, design and performance

3475
 Journal of Theoretical and Applied Information Technology
30th June 2019. Vol.97. No 12
© 2005 – ongoing JATIT & LLS

ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195

[33] Guo, Jing-Ming, Hsia, Chih-Hsien, Liu, Yun- [45] Chang, Kyong, Bowyer, Kevin W., Sarkar,
Fu, et al. Contact-free hand geometry-based Sudeep, et al.Comparison and combination of
identification system. Expert Systems with ear and face images in appearance-based
Applications, Vol. 39, No 14, 2012, pp. 11728- biometrics. IEEE Transactions on pattern
11736. analysis and machine intelligence, Vol. 25, No
[34] Luque-Baena, Rafael M., Elizondo, David, 9, 2003, pp. 1160-1165.
López-Rubio, Ezequiel, et al. Assessment of [46] Kumar, Ajay, Wong, David Cm, Shen, Helen
geometric features for individual identification C., et al.Personal verification using palmprint
and verification in biometric hand and hand geometry biometric. In
systems. Expert systems with applications, Vol. : International Conference on Audio-and
40, No 9, 2013, pp. 3580-3594. Video-Based Biometric Person Authentication.
[35] Goldberg, David E. and Holland, John H. Springer, Berlin, Heidelberg, 2003, pp. 668-
Genetic algorithms and machine 678.
learning. Machine learning, Vol. 3, No 2, [47] Uludag, Umut, Pankanti, Sharath, Prabhakar,
1988, pp. 95-99. Salil, et al.Biometric cryptosystems: issues and
[36] Kumar, Ajay and Zhang, David. Personal challenges. Proceedings of the IEEE, Vol. 92,
recognition using hand shape and texture. IEEE No 6, 2004, pp. 948-960.
Transactions on image processing, Vol. 15, No [48] Jain, Anil K., Flynn, Patrick, and Ross, Arun
8, 2006, pp. 2454-2461. A. (ed.). Handbook of biometrics. Springer
[37] Oden, Cenker, Ercil, Aytul, and Buke, Burak. Science & Business Media, 2007.
Combining implicit polynomials and geometric [49] International Biometric Group (Ibg),
features for hand recognition. Pattern Findbiometrics, Global Identity Management,
Recognition Letters, Vol. 24, No 13, 2003, pp. https://findbiometrics.com/
2145-2152. [50] Prabhakar, Salil, Kittler, Josef, Maltoni,
[38] Shu, Wei and Zhang, David. Palmprint Davide, et al.Introduction to the special issue
verification: an implementation of biometric on biometrics: Progress and directions. IEEE
technology. In : Proceedings. Fourteenth Transactions on Pattern Analysis and Machine
International Conference on Pattern Intelligence, Vol. 29, No 4, 2007, pp. 513-516.
Recognition (Cat. No. 98EX170). IEEE, 1998. [51] Zou, Xuan, Kittler, Josef, and Messer, Kieron.
pp. 219-221. Illumination invariant face recognition: A
[39] Hill, Robert B. Apparatus and method for survey. In : 2007 first IEEE international
identifying individuals through their retinal conference on biometrics: theory, applications,
vasculature patterns. U.S. Patent No and systems. IEEE, 2007, pp. 1-8.
4,109,237, 22 août 1978. [52] Cao, Kaidi, Rong, Yu, Li, Cheng, et al. Pose-
[40] Li, Stan Z. Encyclopedia of Biometrics: I-Z. robust face recognition via deep residual
Springer Science & Business Media, 2009. equivariant mapping. In : Proceedings of the
[41] Kumar, Ajay and Wu, Chenye. Automated IEEE Conference on Computer Vision and
human identification using ear Pattern Recognition. 2018. pp. 5187-5196.
imaging. Pattern Recognition, Vol. 45, No 3, [53] Stern, Guillaume, Fu, Zehua, and Ardabilian,
2012, pp. 956-968. Mohsen. 3D Face Analysis for Healthcare. In
[42] Wang, Liang, Tan, Tieniu, Hu, Weiming, et : Biometrics under Biomedical Considerations.
al. Automatic gait recognition based on Springer, Singapore, 2019, pp. 147-160.
statistical shape analysis. IEEE transactions on [54] Alqahtani, Faleh, Banks, Jasmine, Chandran,
image processing, Vol. 12, No 9, 2003, pp. Vinod, et al. Detection and tracking of faces in
1120-1131. 3D using a stereo camera
[43] Nalwa, Vishvjit S. Automatic on-line signature arrangements. International Journal of
verification. Proceedings of the IEEE, Vol. 85, Machine Learning and Computing, Vol. 9, No
No 2, 1997, pp. 215-239. 1, 2019, pp. 35-43.
[44] Chibelushi, Claude C., Mason, John S., and Nguyen, Anh, Yosinski, Jason, Et Clune, Jeff.
Deravi, R. Integration of acoustic and visual Multifaceted feature visualization: Uncovering
speech for speaker recognition. In : Third the different types of features learned by each
European Conference on Speech neuron in deep neural networks. arXiv preprint
Communication and Technology, 1993. arXiv:1602.03616, 2016.

3476