Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology

ISSN No: - 2456 - 2165

A Review on Unimodal and Multimodal Biometric

Systems
Krishnakumari y, Savitha G,
M.Tech Student, Associate Professor,
Computer Network Engineering, Dept. of CSE,
Dept. of CSE, BNM Institute of Technology
BNM Institute of Technology Bangalore-560070,
Banglore-560070,

Abstract: Biometrics is a method to verify the identity of a essential in biometrics. Trust in these electronic transactions is
person based on psychological characteristics or essential to the healthy growth of the global economy.
behavioral characteristics. Examples of biometrics are Biometrics is integrated with other technologies such as smart
fingerprints, iris, face, ear, voiceprint, palm print, cards, encryption keys and digital signatures. Biometrics is
signature, and handwriting or keystrokes patterns. used in our daily lives. Biometric systems are more accurate
Biometric systems are classified into two types: (1) and convenient for authentication and identification of a
Unimodal biometric system (2) Multimodal biometric person.
system. There is a limited accuracy in the unimodal
biometric system. Accuracy can be improved by using There are three different types of authentication for security.
multimodal biometric system. Two or more biometric They are something you know it provides a PIN, password, or
traits are used in multimodal biometric system. Here we details of persons information, something you have it
present characteristics of biometrics, comparison of includes a smart card, card key or secure ID card and
biometric modalities, difference between unimodal and something you are which provides biometric information.
multimodal biometrics, limitations of unimodal biometrics, Among these, biometric systems provide secure and
fusion levels in biometrics. convenient authentication tool and it is not able to stolen,
barrowed or forgotten and forging biometric information is
Keywords Biometrics, Unimodal biometrics, Multimodal very impossible.
biometrics, Fusion levels, Recognition methods.
II. GENERAL BIOMETRICS TECHNOLOGY
I. INTRODUCTION
A. Processing phases of biometrics
Based on physiological or behavioral characteristics, biometric
systems are used to recognize a person. Some of the examples There are three different processing phases are available in
of biometrics are face, fingerprint, hand geometry, iris, retinal, biometrics: they are (1) Enrollment phase (2) Verification
signature and voice. To obtain highly secure personal phase and (3) Identification phase.
identification and verification details, biometric systems are
widely used. The need of identification and verification Enrollment Phase
technologies is very important as the level of security breaches
and transaction fraud increases. In this phase, template of the individual person images is
stored in the database to check persons identity. Using image
To provide confidential financial transactions and personal of that person, features are extracted.
data privacy, biometrics is used. Biometrics is used in federal,
state and local governments, in military and commercial Verification Phase
applications. And also in secure electronic banking,
government IDs, retail sales, health and social services are In this phase, the person is verified with his template which is
using these technologies. Authentication based applications available in the database by comparing persons captured data.
include network, data protection, remote access to resources,
workstation, transactions and web security. To provide healthy
growth of the global economy, electronic transactions are

IJISRT17MY170 www.ijisrt.com 514

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

Identification Phase requires size that is the captured image is

preprocessed.
In the identification phase, the system will store all users Feature Extraction: In this module, once the image is
details. So the system identifies an individual by searching the captured and preprocessed from the sensor module
templates of all users in the system. The system performs one and preprocessing module respectively. Then
to many comparisons to verify an individual identity, if it is features are extracted from the image using feature
enrolled in the system record. extraction module.
Matching Module: In this module, as we know
B. Working of Biometric system template of the individual data is stored in the
database. So the comparison between the features
The block diagram of biometric system as shown in figure 1
extracted from the input data and the template stored
and its working as follows:
in the database are performed.
Sensor Module: In this module, different sensors are System Database Module: In this phase, templates of
used to capture biometric data of the individual. all users are stored in the database.
Preprocessing: Here, to remove noise from the
image, filters are used. Image is converted into

Figure 1: Block diagram of biometric system

III. MOTIVATION IV. BIOMETRIC MODALITIES

Using biometrics, it is easy to recognize an individual based A biometric modality is a method used to verify or recognize a
on the recognition methods. To recognize an individual it has biometric trait. Examples of biometric traits are face,
some reasons, they include reducing costs and improving fingerprint, hand geometry, palm print, iris, voice, signature,
scalability, reducing error rates, improving accuracy, gait, and keystroke dynamics. The users biometric features
improving convenience and increasing physical safety. To involve both biological and behavioral aspects. Some common
verify a person it is required to have passwords, names, social biometric modalities are summarized here.
security numbers, tokens and PINs, so that the person may
access its services or benefits. For example ATM card and its Face recognition method requires a digital camera to develop
corresponding PIN are required to access an automatic teller a facial image for authentication. It analyzes facial
machine (ATM). characteristics. The casino industry is working on this

IJISRT17MY170 www.ijisrt.com 515

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

technology for quick detection by security personnel to create other biometrics because it enjoys a synergy with existing
a facial database of scam artists. processes but other biometrics does not.

In fingerprint recognition method, there are various Voice authentication is based on voice to print authentication
approaches to fingerprint verification. It is found on a fingertip that is it converts voice into text but then it is not based on
and looks like a pattern on a fingerprint. Some approaches are voice recognition. Voice biometrics does not require any
matching minutiae, others use straight pattern- matching hardware and it has the most potential for growth. And most
devices, and moirefringe patterns and ultrasonics. Some of the computer systems contain a microphone. Voice
verification approaches are used to detect when a live finger is Biometrics has some limitations they are poor quality and
presented and some cannot. ambient noise can affect verification. Voice recognition is not
user friendly because enrollment phase is more complicated
For in-house systems fingerprint verification may be a good compared to other biometrics. Therefore voice authentication
choice. Fingerprint verification can provide adequate method should require improvement.
explanation and training to users and it operates in a controlled
environment. The workstation access applications are based Palm Prints uses the combination of some of the features of
on fingerprints due to low cost, small size and ease of fingerprints and hand geometry. Human palms are larger like
integration. fingerprints and it contains ridges and valleys. Palm prints are
mainly used in the forensic community like fingerprints. And
Hand Geometry recognition method used to analyze and also palm prints can often be found at crime scenes.
measure the shape of the hand. Biometric system is easy to use
and provides good performance characteristics. Biometrics is Gait recognition method includes how a person walks, and
suitable where there are more users available and users can human recognition is based on the distance and period of time.
access the system infrequently. In this hand geometry Gait recognition systems are used to detect the human spatio
recognition method, accuracy is very high if there is desired temporal attributes and it is based on image processing. Some
and flexible performance tuning and configuration. Hand of the factors of gait are includes choice of footwear, the
geometry are available in various scenarios for example time walking surface, and clothing. The development stage is used
and attendance recording, these are proved in organizations. in gait recognition systems.
Hand geometry is the first step in every biometric project due
to ease of integration into other processes and systems and V. BIOMETRICS COMPARISON
coupled with ease of use.
There are various numbers of pros and cons for every
In Iris recognition method, it has colored ring of tissue that biometric system. The main aim of biometrics is to change the
surrounds the pupil to analyze features of iris images. Iris existing password. Biometrics uses both biological and
scanning has higher potential than the average template physiological features to identify a person. Biometrics has
matching performance. For identification phase, Iris some of the features which include iris patterns, retina design,
recognition method is one of the few devices that work well facial geometry, fingerprints, voice recognition and hand
and it works with glasses in place. The disadvantages of iris recognition and so on.
scanning devices are ease of use and system integration.
The following seven factors are:
Retina recognition method used to analyze the layer of blood
vessels located at the back of the eye. This technique uses a Universality
low- intensity light source through an optical coupler to scan Uniqueness
the unique patterns of the retina. Retinal scanning requires the Permanence
user to look into a receptacle and focus on a given point. Measurability
Retinal scanning can be accurate. If you wear glasses, then Performance
retinal scanning is not particularly convenient or these are Acceptability
concerned about having close contact with the reading device. Circumvention
This technique works well even though it is not accepted by
all users. In Universality, using biometric trait every individual should
access the application. In Uniqueness, for every person the
Signature verification method analyzes how a user signs given biometric trait is different from other person. In
her/his name. Some of the signing features are speed, velocity, permanence, for a given matching algorithm, biometric trait
and pressure. These are the finished signatures static shape for a person is invariant over time. The biometric trait which
and very important. Signature verification is different from changes significantly is not a good biometric. In
measurability, the biometric trait uses suitable devices, these

IJISRT17MY170 www.ijisrt.com 516

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

devices should be able to acquire and digitize the trait for elderly and young children, fingerprint images are not able to
every individual and it is inconvenience to the biometric trait. capture properly due to faded fingerprints or underdeveloped
The biometrics system uses acquired raw data to process and fingerprints ridges. And for groups of population it is not
to extract features from the biometric trait. In performance, compatible.
biometrics system should have the higher accuracy to meet the
requirements of the application and to achieve this accuracy it Unimodal system is susceptible to inter class similarities for
requires recognition accuracy and the resources are required to large population. For identical twins facial recognition method
achieve the accuracy. In acceptability, every individual should may not work correctly. Inaccurate matching is the major
have biometric trait in the system and these individuals will issue in the recognition method for identical twins, so camera
use application to present their biometric trait to the system in cannot distinguish between the two subjects. By using
the large population. In circumvention, using biometrics it is unimodal biometrics, the data can be forged or imitated due to
easy to imitate the artifacts using biometric traits for example, spoof attacks. Using rubber fingerprints person information
mimicry can be used for behavioral characteristics and for can be easily spoofed and it is possible in the fingerprint
physiological characteristics fake fingers are used to imitate recognition systems.
the biometric trait of an every person. Security should be very
important, to conform the needs of the application. VIII. MULTIBIOMETRICS

VI. UNIMODAL AND MULTIMODAL Multibiometrics uses two or more biometric traits. By using
BIOMETRICS facial images, fingerprints, iris scanning, hand geometry, voice
recognition methods and so on, they can identify a individual
For identification and verification features, biometric uses a person. To measure two or more biometric characteristics,
single biometric trait of the person is referred as unimodal biometric systems take input from single or multiple sensors.
biometrics. Biometrics which uses more than two biometric Example of multibiometrics is system that combines face and
traits of the individual to identify a person is called as iris characteristics for biometric identification. In order to
multimodal biometrics. The recognition rate can be improved improve the recognition rate, multibiometrics combines the
by using multimodal biometrics. Compared to unimodal two or more biometric modalities. Unimodal biometrics does
biometrics, multimodal biometrics is most widely used in the not provide the desire performance. So the best approach to
organizations. improve accuracy and performance is to use is
multibiometrics.
VII. LIMITATIONS OF UNIMODAL BIOMETRIC
SYSTEMS Multibiometrics has several meanings as shown here:

Biometrics is used in many applications, some of the main Multisensors

applications are border control and voter id issuance. In Multiple algorithms
unimodal biometrics, theoretically it might be very proficient Multiple instances
but in reality it has various numbers of challenges when Multiple samples
enrolling large number of people. The unimodal systems are
Multimodal
not suitable for all applications, this is the major issue with the
unimodal biometrics. Therefore multimodal biometrics is used
In multisensors, single biometric trait is used but then to
to overcome the limits of unimodal biometrics.
capture the biometric data, multiple sensors are used. For
example, to capture different angles on a face multiple
Limits of unimodal biometrics are follows:
cameras are used in the facial recognition system. In multiple
algorithms, using different algorithms captured data are
The biometric trait uses susceptibility to remove noisy or bad
processed. For example, by using minutiae and texture
data. The biometric technology uses the captured biometric
fingerprint images are processed. In this method the hardware
data. And due to imperfect acquisition conditions the
costs and sensors are saved but then complexity is increased.
biometric data might be distorted. By using facial recognition
In multiple instances, for the same modality multiple instances
method the limitations can be seen in applications.
are used in the biometrics. For example, instead of using one
fingerprint image multiple fingerprint images may be matched
With the illumination conditions and facial expressions using
like irises of both eyes. In multisamples, for the same
these, facial images might affect the quality of the facial
biometric trait multiple samples are acquired. For example,
features. Example of fingerprint recognition method is that it
different portions of the fingerprint, it takes multiple images of
leads to false database matching because the scanner is not
this fingerprint and multiple angles of face are captured for the
able to read fingerprints clearly. An imposter leads to falsely
fingerprint biometrics. In multimodal, data can be combined
accepted and enrolled person leads to incorrectly rejected. For

IJISRT17MY170 www.ijisrt.com 517

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

by using different modalities, for example face and fingerprint biometric attributes, biometric fusion is used widely in the
or iris and voice. This biometrics used to capture and process industry.
each modality by using both hardware and software systems.
In biometrics system, implementations of multibiometrics can
IX. FUSION LEVELS IN MULTIBIOMETRICS be done by using levels of fusion. To address number of issues
in the biometrics, fusion is used. Some of the issues are
In multibiometrics, two or more biometric trait is used in robustness, applicability, accuracy, efficiency and universality.
biometric systems and also decision channels used more than To increase robustness of the multibiometrics, various levels
one. Biometric fusion is defined as its aim is to design a of fusion are used for fusing the biometrics traits. Four types
procedure, which combines classification outcome from each of fusions are available they are as follows: sensor level,
biometric channel. To decrease the weakness of individual feature level, matching score level and decision level. The
measurements and to enhance the strengths using different block diagram of fusion levels in a multibiometrics as shown
in figure 2.

Figure 2: Multibiometrics Fusion levels

A. Fusion using sensor level biometric traits are compatible, feature level provides more
Merged biometric trait is formed by using different sensors. accuracy.
For example fingerprint scanner, iris scanner and video
camera etc. Fusing of biometrics traits is done by using C. Fusion using Matching score level
different sensors. And these biometrics traits are processed.
Instead of combining feature vectors, they are processed
B. Fusion using Feature level separately in the biometrics. Then matching score for each
Signals are processed first which are coming from different individual biometric trait is obtained and on the basis of
biometric traits in the feature level. And later from each accuracy of biometrics trait, composite matching score is
biometric trait, separately feature vectors are extracted. In found by fusing matching level. And later classifications are
feature level fusion, signals coming from different biometric used. To combine match scores, number of techniques is used
channels are first processed after which the feature vectors are such as mean fusion, highest rank, logistic regression etc. By
extracted separately from each biometric trait. The fusion using different traits, normalization of scores are acquired
algorithms are used to form composite feature vector by which is the important benefit of this fusion. To achieve this
combining feature vectors and later classifications are applied normalization of match scores, some of the techniques used
for feature vectors. To select useful features, reduction are z-score, piecewise linear, min-max etc. Less complexity is
techniques are used in feature level. Compared to matching obtained from matching score level compared to other fusions
score method, feature level contains richer information of and therefore this level of fusion is used widely.
biometrics and therefore good recognition results are obtained
from feature level fusion. And also when features of different

IJISRT17MY170 www.ijisrt.com 518

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

E. Fusion using Decision level Vincenzo Cont et al., [5] proposes an system which has
Separately pre-classification of each biometric trait is done capacity to reject low-quality items that is for online biometric
first in the decision level. In the biometrics, individual authentication system. Some of the databases are FVC
biometric trait is captured first and extraction of features is databases, CASIA, and BATH are used to contain images with
done from the captured trait. Classification of traits is done as different quality. Those images includes low, medium, high
either accepts or rejects on the basis of extracted features. By quality biometric acquisitions. And also they includes partial
combining the outputs of different traits, final classification of and corrupted images. Due to these reasons biometrics do not
biometrics is done. achieves better results. A unified biometric descriptor is used
and it is working on a template-level fusion algorithm.
X. PREVIOUS WORK Therefore matching algorithm is able to process fingerprint-
codified templates, iris-codified templates, and iris and
Sameer P Patil et al., [1] proposes that for many business fingerprint-fused templates.
processes the reliable personal recognition is critical. Based on
behavioral and physiological characteristics, biometrics used S. Liu et al., [6] proposes a user access, e-commerce and other
to refer to automatic recognition of a person. Many biometric security applications by using secure authentication method.
and non biometric components are used to provide reliable With the help of physically and behavioral characteristics
personal recognition. Biometrics has lot of limitations on biometrics are used to identify a person, for that reason
security of a system. The main two things that security of a biometrics is used. For examples fingerprint, hand or palm
system depends on requirements of an application and cost geometry, and retina, iris, or facial, signature, voice,
benefit analysis. keystroke, pattern, and gait. Here, signature and voice
biometrics traits are used to implement the proposed work.
Sameer P Patil et al., [2] proposes an iris and fingerprint H. B. Kekre et al., [7] proposes biometrics technology for
biometric traits for user verification system. Two different personal authentication. For biometrics system a lot research is
types of fusion levels are used and compared, they are going on in this technology. By using sensors, human traits are
decision level and score level fusion. The methods used for captured with the help of these technologies. Here mainly
iris recognition, to extract region of interest are less complex. discussing about an biometrics domain research for
Better results are obtained from individual modalities. And iris application development, which is focused on interfacing
and fingerprint features are fused to increase the recognition sensor devices. They are used to capture data in usable form.
accuracy of the system by using fusion levels. The component object model and .net platform are used in the
biometrics system. they are used to interface sensors in the
Anil Jain et al., [3] proposes online fingerprint verification for biometrics. There are three components authentication,
the design and implementation of biometrics. It operates in authorization and accountability for security.
two stages. First stage is minutia extraction and second stage
is minutia matching. Minutia extraction method is much faster Gunjan Jiwnani and Nisheeth Saxena [8] proposes
and more accurate. An alignment based elastic matching multibiometrics identification technology by using fuzzy
algorithm is developed for minutia matching algorithm. It has method. In proposing model, it is using two biometric traits
the capability of finding correspondences between input fingerprint and iris and after performing the processing
minutiae and the stored template. And also it has the ability to individually. Fuzzy technique is applied and finally takes
compensate the nonlinear deformations and inexact pose decision that the authentication is very high or low. Biometric
transformations between fingerprints. In this system, the systems are used for authentication due to limitations of
verification accuracy is 99% over a 15% of reject rate. By traditional systems like using PIN, PASSWORD etc. Single
using inkless scanners, it has captured two sets of fingerprint biometric trait also has some limitations so to overcome these
images and tested on biometrics. limitations, biometric systems uses multi biometric
authentication which gives result with more accuracy but also
Anil Jain et al., [4] proposes by using fingerprints for requires more storage as compared to single biometric.
automatic identity authentication system. It uses an alignment-
based elastic matching algorithm. It is capable of finding the Piyush G. Kale, Khandelwal C.S. [9] proposes PCA technique
correspondences between minutiae without resorting to an for IRIS and Finger print biometrics. For proposed model,
exhaustive search. The biometrics achieves an excellent multibiometrics fuses PCA minutia extraction and Weighted
performance. The reason is by using different fingerprints, it LBP feature extraction. The results from these extractions are
has the ability to compensate for the nonlinear deformations applied on different biometric traits. To identify a person the
and inexact transformations. The advantages of biometrics in IRIS and Fingerprint are used in the proposed system. To
this system are response time and accuracy. compare performance and accuracy different recognition
methods are used. Examples of classifiers are SVM & ANN.
These are used for matching.

IJISRT17MY170 www.ijisrt.com 519

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

H B Kekre, V A Bharadi et al., [10] proposes face and iris [8] Gunjan Jiwnani, Mr. Nisheeth Saxena, Multimodal
biometrics for a combination of unimodal and multi- Biometric Authentication using Fingerprint and Iris:
algorithmic systems. By using multilevel decomposition a Review, International journal of computer science
algorithm, face features are extracted, with the help of Kekres & Communication Networks, Vol 5(2), 115-119
wavelet. And by using 1 D transform of row and column [9] Piyush G. Kale, Khandelwal C.S., M.E.(EC), Department
mean, iris features are extracted. Kekres wavelet based of Electronics & Telecommunication, IRIS & Finger
texture features and Kekres Fast Code book Generation Print Recognition Using PCA for Multi Modal
(KFCG) & Kekres Median Codebook Generation (KMCG) Biometric System, International Conference on
algorithms are used for VQ codebooks. Global Trends in Engineering, Technology and
Management (ICGTETM-2016)
XI. CONCLUSION [10] H B Kekre, V A Bharadi, V I Singh, V Kaul, B
Nemade., Hybrid Multimodal Biometric
Here, general biometrics, motivation of using biometric Recognition using Kekres Wavelets, 1D Transforms
system, biometric recognition methods, and numerous issues & Kekres Vector Quantization Algorithms Based
of unimodal biometrics has been discussed. A biometric Feature Extraction of Face & Iris 2nd International
system which uses single biometric trait to identify a person it Conference and workshop on Emerging Trends in
is referred to as unimodal biometrics. Using unimodal Technology(ICWET-2011)
biometrics it is very difficult to achieve performance. So in [11] Kamel Aizi, Mohamed Ouslim, Ahmed Sabri., Remote
order to improve performance of the biometrics, multimodal Multimodal Biometric Identification Based on the
biometrics is used. An overview of applications of Fusion of the Iris and the Fingerprint, International
multibiometrics and its different fusion levels are discussed. In Conference on Information Science and control
todays applications, higher performance and level of security Engineering (ICISCE), pp. 1-5, 2015
can be improved due to multibiometrics. [12] B. C. Kovoor, M.H. Supriya and K. Poulose Jac, A
prototype for a multimodal biometric security system
REFERENCES based on face and audio signatures, International
Journal of computer Science and Communication
[1] Sameer P Patile, Tushar N Raka, Shreyas O Sarode, vol. 2, no. 1, pp. 143-147, January-June 2011.
Multimodal Biometric Identification System: Fusion [13] Zatin Singhal, Preeti Gupta and Kavitha Garg,
of Iris and Fingerprint, International Journal of Biometric recognition Personal identification
Intelligent Information Technology technique, Ijcem International Journal of
Application (IJIITA), Vol. 2, No.6, ISSN 1999- Computational Engineering & Management, vol. 15,
2459 (print), December,pp. 279-285, 2014. Issue 3, May 2012.
[2] Sameer P Patil, Tushar N Raka, Multimodal [14] J.A. Unar, Woo Chaw Seng and Almas Abbasi, A
Biometric Identification System: Fusion of Iris and review of biometric technology along with trends and
Fingerprint. prospects, Pattern recognition, accepted manuscript.
[3] Anil Jain, Lin Hong, On-line Fingerprint [15] A. H. Mir, S. Rubab and Z. A. Jhat, Biometrics
Verification International Journal of Advanced verification: a literature survey, International
Research in Computer Science and Technology , Journal of Computing and Ict research, vol. 5, no 2,
vol. 2, Issue Special 1, Jan-March 2012. December 2011.
[4] Anil Jain , Lin Hong , Sharath Pankanti, An Identity [16] Rohit Katiyar, Vinay Kumar Pathak and K. V. Arya,
Authentication System Using Fingerprints. A study on existing gait biometrics approaches and
[5] Vincenzo Conti, Carmelo Militello, Filippo Sorbello, challenges, Ijcsi International Journal of Computer
A frequency-based Approach for features fusion in Science issues, vol. 10, issue 1, no 1, January 2013.
fingerprint and iris multimodal biometric [17] M.J. Sudhamani, M.K. Venkatesha, K.R. Radhika,
identitification systems International Journal of Revisiting Feature level and Score level Fusion
Ophthalmol, vol. 8, No.1, Feb 2013. Techniques in Multimodal Biometrics System,
[6] S. Liu, M. Silverman, A practical guide to biometric Proceedings of International Conference on
security technology, IT Professional, Vol. 3, No. 1., Multimedia Computing and Systems (ICMCS), pp.
pp. 27-32, Aug 2002. 881-885, 2012.
[7] H. B. Kekre, V. A. Bharadi, Using Component [18] Y. Tong, F. W. Wheeler, X. Liu, Improving Biometric
Object Model for Interfacing Biometric Sensors to Identification through Quality based Face and
Capture Multidimensional features , International Fingerprint Biometric Fusion, Proceedings of IEEE
Journal of Intelligent Information Technology Computer Society Conference Vision and Pattern
Application (IJIITA), Vol. 2, No.6, ISSN 1999-2459 Recognition Workshops, pp. 53-60, 2010.
(print), December, pp. 279-285, 2009.

IJISRT17MY170 www.ijisrt.com 520

Volume 2, Issue 5, May 2017 International Journal of Innovative Science and Research Technology
ISSN No: - 2456 - 2165

[19] A. Kumar, Y. Zhou, Human Identification Using

Finger Images, IEEE Transactions on Image
Processing, vol. 21, n. 4, pp. 2228-2244 2012.

IJISRT17MY170 www.ijisrt.com 521