International Journal of Scholarly Research (IJSR)

Vol-1, Issue-3, 2017

http://www.scholars.org.in

Implementation of Unimodal to Multimodal Biometric Feature Level

Fusion of Combining Face Iris and Ear in Multi-Modal Biometric System"

Gauravweni S. Hedau
Departement of computer science and engg,
G.H.Raisoni Academy of engg. & technology, Nagpur
ghedau1373@gmail.com

Abstract: In this paper is on combining the features face ear cues). A system that consolidates the evidence presented by
and iris to single modal to multimodal combine of the obtain multiple biometric cues is known as a multi-biometric system.
better accuracy improvement as compared to the privies Multi-sensor in which, more than one sensors are used to
system. The paper discusses a new research area to introduce capture information from the presented biometric trait (e.g.,
ear as the feature for multimodal biometric our system is to capacitive and optical sensors for fingerprints). Multi-sample
implement a multimodal an multialgorithmic biometric system when more than one recording of the biometric trait is used
combining Ear iris and Face recognition. We have developed (e.g., multiple face images can be used for creating the
the secure facial features based authentication system Multi template). Multi-algorithmic where the same biometric data
modal biometric system is one of the major areas of study is processed through multiple algorithms (e.g., minutiae and
identified with large applications in recognition system. Single texture based features for fingerprints). Multi-unit or multi-
modal-biometric systems have to challenge with a variety of instance in which, multiple instances of the same biometric
problems such as noisy data, intra-class variations, non trait are used (e.g., information from images of left and right
universality and unacceptable error rates. Some of these irises is combined) Multi-modal when more than one
limitations can be solved with multi modal Biometric systems. biometric traits are used (e.g., a combination of iris and face).
In this paper we are going implement a multi modal biometric The problem of consolidation of information presented by
system combining the feature of face iris and ear modalities to multiple biometric sources or cues from any of the types
obtain better accuracy as compared to the privies based mentioned above is known as information fusion. The
system. information fusion in a biometric system can be carried out at
Keyword: Multi-biometrics, Biometrics, Fusion level, Feature different levels. Sensor Level Information coming from
extraction, PCA, SVM. different sensors is combined. Feature Level The biometric
information extracted in form of features is combined. Score
I. INTRODUCTION Level Match scores of individual biometric comparisons are
Biometric recognition technology can be widely used in the combined. Decision Level The results of individual
field of authentication, security, human computer interaction biometric comparisons are combined.
system, criminal identification and investigation, security. Rank Level When the output of each biometric system is a
Unimodal biometric recognition system variety of problem subset of possible matches (i.e., identities) sorted in decreasing
such as the problem to solved can be overcome by employing order of confidence, the fusion can be done at the rank level.
multimodal biometric system used This is relevant in an identification system where a rank may
be assigned to the top matching identities.
Biometrics: Overall multi-modal biometrics system based on face, iris and
Automated recognition of individuals based on their ear could their connection relationship of physiological and
behavioural and biological characteristics is called biometrics. biological location. It would significantly improve accuracy.
Some examples of biometric characteristics are fingerprint, Multi-modal biometrics based on face and ear is expected to
iris, face (2D and 3D), retina, palm print, hand veins, ear, make a breakthrough in the field of biometrics In this paper we
knuckles, DNA, voice, signature, gait, typing patterns, etc. first the overcome the disadvantage unimodal to multimode
These characteristics are denoted as biometric traits or biometric system. And the combining multimodal biometric
modalities. Since the biometric traits are intrinsically bound to based on face and ear and the discuss new research area
the person, they can be used to establish his identity with high .finally find to better accuracy in this paper.
degree of confidence.
II. RELATED WORK
Multi-biometrics:
An important development in the field of biometrics is to In this section we analysis previously done research work on
combine information from multiple biometric sources (i.e., biometrics system and there different types of biometrics as

International Journal of Scholarly Research Page 16

International Journal of Scholarly Research (IJSR)
Vol-1, Issue-3, 2017
http://www.scholars.org.in

well as previously work done on multi-modal biometric been existing using faces ears and iris modalities for self
systems with different algorithms and design. The produced databases. Multimodal biometrics has resulted
summarizing review of earlier done related work is as follows. enhanced performance in provisions of recognition
Kirti V. Awalkar, Sanjay G. Kanade, Dattatray V. Jadhav, accurateness, FAR and FRR.
Pawan K. Ajmera [1] have build a multi-modal and multi R. Brunelli and D. Falavigna [8] have proposed Multimodal
algorithmic biometric system by combining iris and face. The systems combine the verification presented by different body
iris features are extracted using the Daugmans algorithm character for creating and establishing identity. They has
which results in binary features. For face have used two types proposed a multimodal biometric systems utilized face and
of Features: Gabor filters based and Local Binary Patterns voice features to devlop the identity of an individual. actually
(LBP) based. They have performed broad experiments to uncorrected traits (e.g., ngerprint and iris) are expected to
estimate the system on widely available databases using result in better development in performance than associated
regular well defined protocols. traits (e.g. voice and lip movement).
Lin Zhang, Lida Li, Hongyu Li, and Meng Yang[2] have
proposed a new method for 3D ear identification, that is
LCKSVD_LHST. There donations are mainly from two III. MULTIMODAL BIOMETRIC FOR FEATURE
aspects. They have get used to LC-KSVD, a state-of-the-art LEVEL FUSION FACE IRIS AND EAR TO
model for supervised dictionary learning, to the relevance of COMBINE UNIMODAL
3D ear recognition. Second aspect is for feature extraction, The input images it first given to Viola Jones detection where
they have proposed an approach based on local histograms of the face extraction if face are not found when a new images
STs, which is quite helpful and strong to small placement taken at the input else the face is the process wired created
errors. They have achieved higher recognition rate than the object detection to find out the eyes of the input face. Once the
other competitors evaluated. And apart from all this its eyes are extracted then we will use facial geometry to find out
computational difficulty is extremely low at the test stage, the electronic of eyes, face, fingerprint, etc. Once face, eyes
manufacturing it quite appropriate for large-scale identification and ear are extracted then to find out of feature of the face,
applications. eyes and ear. The find out mainly the structure property of all
Gandhimathi Amirthalingam, Radhamani.[3] have projected the components .For any new images SVM classified apply in
framework of the multimodal biometric system using face and order to get the classification images.
ear. They have also discusses the levels of fusion that are
promising and appreciate the types of difficulties focused by
previous research work in this area.
Javier Galbally, Sbastien Marcel, Julian Fierrez[4] have
present a new software-based fake detection method that can
be used in multiple biometric systems to detect different types
of fraudulent unauthorized access attempts which is made by
third party for wanting the access of some precious system
information or some other important thing. They have set the
objective of proposed system is to improve the security of
biometric recognition architecture, by adding liveness
evaluation in a fast, user friendly, and non-intrusive manner,
through the use of image quality assessment.
Teddy Ko[5] have overviews and discusses the various
scenarios that are possible in multimodal biometric systems
using fingerprint, face and iris recognition, the levels of fusion
that are possible and the integration strategies that can be
adopted to fuse information and improve overall system
accuracy.
A.A. Darwish, R. Abd Elghafar and A. Fawzi Ali [6] have
present study has designed to build up a multimodal biometric
system for personal identification. Investigational results have
shown that integrated face and ear recognition system. He has
implemented this system at different database. At the end he
got result and concludes that combining face and ear is a good Fig: Flow diagram of system modal
method because it offers a high precision and security. In our proposed methodology there are major six step are
Snehlata Barde, A. S. Zadgaonkar, G. R. Sinha[7]have shows, to be followed:
PCA, Hamming Distance based multimodal biometrics has

International Journal of Scholarly Research Page 17

International Journal of Scholarly Research (IJSR)
Vol-1, Issue-3, 2017
http://www.scholars.org.in

1. Created the database: first the created database for of enter of of

the input images saved the database for 100 person in images correct (C)/(B)
enters and more than enteric. database extracted output
2. Pre-processing tested (C)
Pre-processing the second step is the pre-processing (B)
Detected the face path and then apply control point and 5 3 2 2/3*100= 66.66%
encoding the face path. The control point and the 10 10 8 8/10*100=80%
boundary box to used the PCA and histogram 20 20 17 17/20*10=85%
algrothium 30 30 27 27/30*100=90%
40 40 37 37/40*100=92.5%
50 50 47 47/50*100=94%
60 60 57 57/60*100=95%
70 70 66 66/70*100=98.57%
80 80 79 79/80*100=98.75%
90 90 89 89/90*100=98.88%
100 100 99 99/100*100=99%

Table 1: Face, Iris and Ear

3. PCA: PCA is one of the most popular face recognition

algrothium is used the database for the purpose the Number of Number of Number Accuracy
work contain train subject and test subject test enter in images of correct
biometric used face and ear .first the input images to database extracted output (C)/(B)
face presented than pre-pressing detected face path for tested (C)
control point to images to encoding face for histogram (B)
algrothium to detected face using the PCA feature. 10 10 8 8/10*100=80%
20 20 17 17/20*100=85%
30 30 27 27/30*100=90%
40 40 38 38/40*100=95%
50 50 48 48/50*100=96%
60 60 58 58/60*100=96.66%
4. Created GUI: 70 70 69 69/70*100=98%
80 80 79 79/80*100=98.75%
Graphic user interface created and the created database for
matching person identification and evaluation database face Table 2: face and iris
and ear. combine mapping and person name to save to the
database To the feature level combine fusion face ,eyes and
ear. input the 10 images and the testing to find corrected
output.
Face iris and Ear
5 .matching: decision is the matching verification of the person 120
for multimodal biometric system various system to used and
100
various algrothium to use PCA, for feature extraction, viola
Jones face detect SVM classification used. first to the input 80
images than images extracted tested and than corrected output 60
and than to find accuracy to compare the privies system and Face iris and
unimodal to find the accuracy to implementation of the better 40 Ear
accuracy find out of them. 20
0
IV. Experimental and results setup:
5 20 40 60 80 100

Number Number Number Accuracy

Fig 1: face iris and ear

International Journal of Scholarly Research Page 18

International Journal of Scholarly Research (IJSR)
Vol-1, Issue-3, 2017
http://www.scholars.org.in

Recognition, IEEE TRANSACTIONS ON IMAGE

PROCESSING, VOL. 23, NO. 2, FEBRUARY 2014.
Face and Iris [6] Teddy Ko, Multimodal Biometric Identification for
Large User Population Using Fingerprint, Face and
120 Iris Recognition, Proceedings of the 34th Applied
100 Imagery and Pattern Recognition Workshop
(AIPR05) 0-7695-2479-6/05 $20.00 2005 IEEE.
80 [7] A.A. Darwish, R. Abd Elghafar and A. Fawzi Ali,
60 Multimodal Face and Ear Images, Journal of
40 Face and Iris Computer Science 5 (5): 374-379, 2009.
20
[8] Snehlata Barde, A. S. Zadgaonkar, G. R. Sinha,
0 Multimodal Biometrics using Face, Ear and Iris
10 20 30 40 50 60 70 80 Modalities, International Journal of Computer
Applications (0975 8887) Recent Advances in
Information Technology, 2014.
[9] R. Brunelli and D. Falavigna, Person identification
Fig 2: face and iris using multiple cues, IEEE Transactions on Pattern
Analysis and Machine Intelligence,vol. 17, no. 10, pp.
955966, 1995.
V. CONCLUSION [10] Z. Zhang, R. Wang, K. Pan, S. Li, and P. Zhang,
Fusion of near infrared face and iris biometrics, in
Advances in Biometrics, Lecture Notes in Computer
Overall accuracy of Ear -Face facial biometric authentication Science,, 2007, pp. 172180.
system. Will be evaluated and tested under various facial [11] D. Maturana, D. Mery, and A. Soto, Face
images. Developed an various algorithm which combines recognition with local binary patterns, spatial pyramid
features from human Iris, Ear and Face for person verification. histograms and naive bayes nearest neighbour
.We achieved significant improvement in the verification classification, in Proceedings of the 2009
performance. improved recognition accuracy compared to International Conference of the Chilean Computer
unimodal and privies system Science Society, 2009.
[12] A. K. Jain, P. Flynn, and A. Ross, Eds., Handbook of
Biometrics. Springer, 2008.
VI. REFERENCES [13] ISO/IEC CD 2382.37, Information processing
systems Vocabulary Part 37 : Harmonized Biometric
[1] Kirti V. Awalkar, Sanjay G. Kanade, Dattatray V. Vocabulary, 2010.
Jadhav, TSSMs BSCOER, Narhe, Pune, [14] R. Brunelli and D. Falavigna, Person identification
Maharashtra, India. A Multi-modal and Multi- using multiple cues, IEEE Transactions on Pattern
algorithmic Biometric System Combining Iris and Analysis and Machine Intelligence,vol. 17, no. 10, pp.
Face 2015 IEEE International Conference on 955966, 1995.
Information Processing (ICIP) Vishwakarma Institute
of Technology. Dec 16-19, IEEE 2015. [15] B. Son and Y. Lee, Biometric authentication system
[2] Lin Zhang, Lida Li, Hongyu Li, and Meng Yang, 3D using reduced joint feature vector of iris and face, in
Ear Identification Using Block-Wise 6th International Conference on Audio and Video-
[3] Statistics-Based Features and LC-KSVD, IEEE Based Biometric Person Authentication (AVBPA03),
TRANSACTIONS ON MULTIMEDIA, VOL. 18, 2003.
NO. 8, AUGUST 2016.
[16] C. H. Chen and C. T. Chu, Fusion of face and iris
[4] Gandhimathi Amirthalingam, Radhamani. G, A features for multimodal biometrics, in Advances in
Multimodal Approach for Face and Ear Biometric Biometrics, Lecture Notes in Computer Science,,
System, IJCSI International Journal of Computer 2005, pp. 571580.
Science Issues, Vol. 10, Issue 5, No 2, September [17] Y. Wang, T. Tan, and A. K. Jain, Combining face
2013. and iris biometrics for identity verication, in 4th
[5] Javier Galbally, Sbastien Marcel, Julian Fierrez International Conference on Audio and Video-Based
,Image Quality Assessment for Fake Biometric Biometric Person Authentication (AVBPA03), 2003,
Detection: Application to Iris, Fingerprint, and Face pp. 805813

International Journal of Scholarly Research Page 19

International Journal of Scholarly Research (IJSR)
Vol-1, Issue-3, 2017
http://www.scholars.org.in

[18] A. Rattani and M. Tistarelli, Robust multi-modal

and multi-unit feature level fusion of face and iris
biometrics, in Advances in Biometrics, Lecture
Notes in Computer Science, 2009, pp. 960969.
[19] B. Son and Y. Lee, Biometric authentication system
using reduced joint feature vector of iris and face, in
6th International Conference on Audioand Video-
Based Biometric Person Authentication (AVBPA03),
2003.
[20] M. Eskandari and O. Toygar, Fusion of face and iris
biometrics using local and global feature extraction
methods, Signal, Image and Video Processing, vol.
8, no. 6, pp. 9951006, 2014.
[21] A. Jain, K. Nandakumar, and A. Ross, Score
Normalization in Multi- modal Biometric Systems,
Pattern Recognition, vol. 38, pp. 2270 2285, 2005

[22] T. Ahonen, A. Hamind, and M. Pietikainen, Face

description with local binary patterns: application to
face recognition, IEEE Transactions on Pattern
Analysis and Machine Intelligence, vol. 28, no. 12,
pp. 2037 2041, 2006

[23] A. A. Ross, K. Nandakumar, and A. K. Jain,

Handbook of Multibio metrics, ser. International
Series on Biometrics. Springer, 2006.

International Journal of Scholarly Research Page 20