2019 International Conference on Electrical, Computer and Communication Engineering (ECCE), 7-9 February, 2019

Design a Secure Voting System Using Smart Card

and Iris Recognition
Md. Mahiuddin
Department of Computer Science and Engineering
International Islamic University Chittagong (IIUC)
Chittagong, Bangladesh
mmuict@gmail.com

Abstract— Bangladesh is one of the countries in which

introduced voting system in parliamentary and assembly polls.
But in every election, the election commission is facing a lot of
troubles and various types of problems throughout the election.
The most familiar issue faced by the election commission is Fig. 1. Traditional Paper Voting System
improper confirmation with respect to the arrangement of
casting the votes, duplication or illegal casting of votes. In this
paper, a secure and new voting system is developed to improve
the existing voting system using smartcard and iris recognition.
Iris is one of the most secure biometric of person identification.
The main goal of this article is to avoid the duplication of
casting votes.

Keywords— Iris, Smartcard, Voting system.

Fig. 2. Electronic voting system
I. INTRODUCTION
Security is the main concern of existing voting systems.
Voting system is a government selection process in every Sometimes an unauthorized person gives vote. Some
democratic nation. Democracy is meant to allow people to politicians try to follow illegal method to win the election. In
vote freely and voting is the right of every people of a paper ballot and EVM systems needed more manpower.
democratic nation. The democratic government depends on These existing systems are much more time consuming and
the results of the election. also slow. In proposed system, we use Irish pattern and smart
Today’s world is the era of internet and computer card. Hence the proposed voting system is more secure than
technology. The uses of computer, internet and electronics existing system.
are increasing day by day. The security system is also Smart card is a card in which a microprocessor and a
upgrading day by day. It should be necessary to upgrade the memory chip are attached used for processing and storing
traditional voting system and also security. The main information respectively. Secure exchange between the
objective of this article is to develop a new idea about voting reader and the card is performed in the card more easily.
system and also ensure the security of it. Smart card has the capacity of store and access data. It also
To ensure the security of voting system, we use iris provides an immediate exchange of necessary information.
recognition in proposed system. Iris recognition is used We can store a person’s iris data and personal information in
because it is highly unique, stable, cannot be duplicated and smart card.
easily captured. In the proposed voting system, the voter identity card is
Different types of voting system have been used around replaced by smart card in which all the details of the person
the world. Paper ballot voting system is an old and unsecure are updated. Only the specified person can poll using their
voting system where it is possible to cast multiple votes from smart card. Here the smart card reader reads the smart card
the same voter. It runs to cast the vote with a ballot paper and and the details of that person is displayed, and then it asks for
a stamp (Shown in Fig.1). verification which is iris recognition. If the iris pattern
matches then the person can poll. The person is allowed to
In electronic voting system (EVM), the process of poll once using by smart card. After voting, if the person
election data is recorded, stored and preceded as digital tries to use smart card again, the smart card reader will
information [1]. Electronic voting device is used as casting access the cards but it will show the message that the person
as well as counting of votes. The block diagram of EVM is has already voted.
depicted in Fig. 2. Design of EVM based on solar power is
reported in [2]. Implementation and formal analysis of A. Hough Transformation
electronic voting protocol using AVISPA is presented
elaborately in [3]. Supeno Djanali et al. proposed a method We use the Circular Hough Transform algorithm to
to secure voting data based on SHA256, digital signature, segment the image. In our proposed system, we find the iris
and RSA asymmetric encryption [4]. A Low cost, low error and pupil boundaries from an eye image using Circular
rate and efficient android based e-voting system is developed Hough Transformation algorithm. It is also used to localize
for the election of EEPIS BEM president [5]. the iris where the pupil boundary and iris boundary are

978-1-5386-9111-3/19/$31.00 ©2019 IEEE

present. Any lines and circles can be identified from an D. Wavelet encoding
image using Circular Hough transformation algorithm. The data in the iris region can be decomposed by
A circle equation with radius r and center (a, b) is wavelets into components and it appears in different
described as follows. resolution. The wavelets are localized in both frequency and
time [9].
(x − a )2 + ( y − b )2 = r 2
E. Gabor Filters
A circle can also be expressed by the equations as Gabor filter can provide optimum combined
follows. representation of a signal in space and spatial frequency. It is
x = a + r sin(θ ) created from a Gaussian kernel function modulated by a
sine/cosine wave. The localization in frequency of a sine
y = b + r sin(θ ) wave is perfect, but it is not localized in space at all. When a
sine wave is multiplied by Gaussian function, it provides
Here use the Hough transform to find the triple localization in space with some loss of frequency
parameters of (a, b, r) to determine the point (xi, yi) [6]. localization. It provides localization in both frequency and
space. Gabor filter has a real part and an imaginary part. The
B. Daugman’s Integro-differential Operator real part is achieved from a cosine modulated by a Gaussian
The pupil region and circular iris are detected by Integro- function and the imaginary part is achieved from modulating
differential operator in Daugman Iris detection a sine wave with a Gaussian function [10].
implementation. The eyelids and eyelashes can also be Daugman’s phase quantization and demodulation method
detected by it. The equation of integro-differential operator is can be represented as follows.
as follows [7].
h{Re,lm} = sgn{Re, lm} {l(ρ,φ)e−iω(θ0 −φ) .e−(r0 −ρ)
2
/ a2
∂ I ( x, y ) .
max ( r , x p , y 0 ) Gσ (r ) *
∂r  r , x0 , y 0
2πr
ds ρφ
2 2
.e − (θ 0 −φ ) /β
ρdρdφ }
Where, I(x,y) represents the image of eye, radius Here, I(ρ,ϕ) is the normalized image of iris, α and β are
described by r, Gσ(r) denotes the Gaussian smoothing 2D wavelet size parameters, (r,θ) represents the polar
function, contour of the circle given by (r, x0, y0) is s. coordinates and ω is wavelet frequency.

C. Normalization F. Log-Gabor Filter

Normalization is performed using the rubber sheet model The relationship between the number of samples and the
proposed by Daugman. In this technique, every point in the frequency and space resolution is disproportional since more
segmented iris region is converted into Cartesian coordinates samples give the higher frequency information and lower
(x,y) to polar coordinates (r,θ). space resolution. The quality of a filter depends on obtaining
the maximum frequency information given a set of time
resolution [11]. From this perspective, Gabor filter is a good
filter because it provides excellent localization on space
resolution and frequency information. However, at certain
bandwidth, a Gabor filter contains non-zero DC component
which means that the response of the filter is depending on
the signal mean value. On the other hand, non-zero DC
Fig. 3. Dougman’s rubber sheet model. component can be acquired for any bandwidth by using Log-
Gabor filter.
The transformation is described by the following The frequency response of a Log-Gabor filter is given as
equation. follows.
I ( x(r ,θ ), y (r ,θ )) → I (r ,θ )
   f  2 
Here,  −  log  
   f0   
x( r ,θ ) = (1 − r ) x p (θ ) + rxl (θ ) G ( f ) = exp 2 
 2 log σ   
y ( r ,θ ) = (1 − r ) y p (θ ) + ryl (θ )    f0   
Where, I(x,y) represent the region of iris, Cartesian Where, f0 is the center frequency, σ represents the filter
coordinates is described by (x,y), corresponding normalized bandwidth and f is the original frequency.
polar coordinates can be represented by (r,θ), and the
coordinates of the iris boundaries and the pupil in the θ G. Hamming distance
direction are represented by (xl, yl) and (xp, yp) respectively Hamming distance (HD) measures the number of
[8]. different bits in two strings of the same length. In another
way, it measures the number of bit shifting is required to
match two strings. It is also used to count the error rate in If one person gives votes in one place the database will
matching. get updated globally that this person has given the vote. If
the voter will go some other place than after scanning the eye
If hamming distance is 0 between two templates, it it will be shown that this person has already given the vote.
represents a perfect match. If hamming distance is 0.5 As a result, the person is not allowed to vote. By this way,
between two templates then the two templates is we can eliminate the chances of registering the person in
independent. A threshold is set to decide the two templates more than one place.
are from the same person or different persons [11].
There is no need to mark voter’s forefinger by the
The benefit of Hamming distance is fast matching speed inedible ink. The database will be updated when a person
because the templates are in binary format. gives the vote. It is more secure than the traditional ballot
Let there are two patterns and N is the total number of paper voting system.
bits present in both patterns. HD is counted by doing Our proposed system is very much time effective and
exclusive-OR (XOR) of the bits of two patterns. So, the fast. Require less man power than ballot paper voting system
equation for Hamming Distance is expressed as follows. and electrical voting system.
1 N
HD =
N
 j =1 X j ( XOR)Y j A. Iris recognition
Iris recognition system consists of five stages, such as,
Evaluation of Hamming distance is performed with bits image acquisition, segmentation, normalization, feature
that generated from the actual iris region. extraction and matching.
Firstly, an iris image is collected from open source
II. PROPOSED VOTING SYSTEM database, CASIA database. In segmentation process, we use
Our proposed voting system is depicted in Fig. 4. Firstly, circular Hough transform and canny edge detector. In
we check the iris database from the smart card. On that time, normalization, we use Doughman rubber sheet model. In
the voting record of a voter is also checked. If the voting feature extraction process, we use 1D log Gabor wavelet.
record shows the voter gives vote then he/she cannot allow And lastly, for matching, we use hamming distance.
for voting. If the voting record shows that the voter cannot
give vote then capture the voter iris image by using iris
scanner camera. Match the captured iris image and smart
card iris image database using hamming distance. If the iris
image is not matched to the smart card database then stop the
process. On the other hand, if the iris image is matched to
the smart card database then allow the voter to give a vote
and update the voting record of the voter.

Fig. 5. Steps of iris recognition systems.

B. Image Acquisition
Iris image cannot be acquired by the normal camera
because it is small in size. The detail of iris cannot visible in
normal camera. There are also light reflection is present
when capturing iris image by normal camera. So, it is
difficult to acquire good images.
We are using “CASIA (Institute of Automation, Chinese
Academy of Sciences) Iris Image Database version 4.0” and
some iris image taken by mobile for image acquisition.

C. Segmentation
For segmentation, Hough transform is better than
Integro-Differential Operator. The errors of segmentation by
Hough Transform technique show that the segmentation is
lower than Integro-Differential Operator [12]. So, we use
Fig. 4. Block diagram of proposed voting system. Hough transform for segmentation.
The chances of voting illegal voting will be reduced. If The edge detection technique is used before Hough
one person takes other person smart card then after scanning transform is applied. For this reason, we choose the ‘canny
the eyes the details of the voters come in front of the officer edge’ to extract the image. We find all the edges in the iris
or poll worker. The officer or poll worker will match that image. In this method, the inner and outer boundary of an iris
voter smart card details from the database. If it doesn’t match is detected (shown in fig. 6).
then the officer or poll worker will get to know that the We use Canny edge detector because it is better than
person is a real person or fake person. In this way, we can other edge detector like Sobel edge detector for iris detection
find the illegal voter and exclude the voter. [13].After finding edge detection point, at each edge point
draw a circle with center at the point with the needed radius.
In segmentation process, the circle is drawn in the parameter
space. So, our x axis is the a-value and the y axis is the b
value and the z axis is the radius (r).
In this way, the inner and outer boundary of iris can be
detected by circular Hough transform. The result of
segmented iris image is shown in Fig. 7.

Original image Normalized Image

Fig. 8. Segmented iris image to normalized iris image

Original image After canny edge detector

Fig. 9. Binary Mask normalized image
Fig. 6. Original image and segented Image

F. Matching
For matching, we use the hamming distance. Hamming
distance of two templates is calculated by shifting one
template left and right bit-wise and a number of Hamming
distance values are calculated from successive shifts.
Illustration of one shifting process is shown in Fig. 10. One
shift is defined as one shift left, and one shift right of a
reference template.

Original image After canny edge detector

Fig. 7. Implementation of Hough transformation

D. Iris normalization
For normalization, we use Doughman’s rubber sheet
model. In this technique, every point in the segmented iris
region is converted into Cartesian coordinates to polar
coordinates. The result of normalization iris image is given
in Fig. 8.

E. Feature extraction
A template representing iris pattern information is
created using a 1D log-Gabor wavelet in the feature
extraction. Fig. 10. Shifting process

Feature extraction is implemented by the normalized iris

pattern is convoluted with 1D Log-Gabor wavelet. First 2D III. RESULTS AND DISCUSSIONS
normalized iris pattern is broken up into a number of 1D
signal and then Gabor filter is used to those 1D signals. The A. Segmented result
rows of the 2D normalized pattern are taken as the 1D signal. The CASIA database image is successfully segmented
Each row corresponds to a circular ring on the iris region accurately but some of iris image cannot be segmented
[14] (Depicted in Fig. 9). properly. The result is shown in Table I.
The encoding process produces a bitwise template Every image of the database has different segmentation
containing a number of bits of information, and a value. The segmentation value of the iris image based on
corresponding noise mask which corresponds to corrupt their circle iris and circle pupil is achieved by using Hough
areas within the iris pattern, and marks bits in the template as transform. The average of the circle pupil and circle iris of
corrupt. The result is given Fig. 9. CASIA database is given in Table II.
 B. Matching Result
In this paper, we use ideal iris for experiment. So, iris
pattern is important for iris recognition. In our experiment,
some image is perfectly identified and some images are not
identified. Some same iris image is not identified perfectly.
The details result is described in Table III.
We use 200 images of 100 person of CASIA iris database
which is ideal iris image. In CASIA iris image, there is no
light reflection, and iris and pupil is visible. The detail of iris
is completely visible. So, the segmentation of iris image of
CASIA iris database is perfectly done. The iris boundary
(iris-sclera) and pupil boundary (pupil-iris) is perfectly
segmented. Fig. 11. Average of central pupil and Average of central iris

TABLE I. SEGMENTATION RESULT OF EYE IMAGE

TABLE III. MATCHING RESULT OF INPUT IMAGE
Number
Number Hamming Successfully Error in
Eye Number of of Error in Image 1 Image 2
of distance recognized matching
Position eye images segmented Segmentation
Persons
eye
1 L 10 10 0
R 10 10 0
2 L 10 10 0 0.4085 Different iris 0
R 10 10 0
3 L 01 10 0
R 07 10 0
4 L 10 10 0
R 10 10 0
5 L 06 02 4
0.3729 Different iris 0
R 01 01 0
6 L 01 10 0
R 08 10 0
7 L 09 10 0
R 10 10 0
8 L 07 07 0
R 07 07 2 0.4853 Different iris 0
9 L 01 10 0
R 05 10 0
10 L 10 10 0
R 10 10 0

0.4003 Different iris 0

TABLE II. AVERAGE SEGMENTATION VALUE OF IRIS IMAGE

CASIA database Image

Iris Avera Average
Image ge of of
Central iris Central pupil
centra central 0 Same iris 0
l iris pupil

115 158 110 127.666 110.33333

118 156 57 6667 33

0 Same iris 0

170 155 113 118.66666

167 154 35 146 67

C. Limitations
We are using 20 non-ideal iris images. The image is
133.33333 captured by us by using smart phone. The pupil and iris are
125 168 110 127 163 48 185 33 not properly identified. There is light reflection of our
database. The details of iris and pupil of our database is not
completely visible. So, the iris image of our database is not
perfectly segmented.
 IV. CONCLUSION [8] Caroline Houston, ” Iris Segmentation and Recognition Using
Circular Hough Transform and Wavelet Features”, Rochester Institute
With the increasing the population day by day, the of Technology.
improvement of voting system is necessary. Undoubtedly the [9] Richard Yew Fatt Ng,Yong Haur Tay,Kai Ming Mok,” A Review of
proposed voting system is techniques are especially good. Iris Recognition Algorithms”, 2008 International Symposium on
We have used iris recognition and smart card for improving Information Technology, Volume: 2, ITSim 2008.
this system. Many bio metric methods are available but iris [10] Libor Masek,” Recognition of Human Iris Patterns for Biometric
recognition has high accuracy rate. Using the smart card, it is Identification”, The University of Western Australia, 2003.
likely to poll from any polling booth rather than the [11] K. Seetharaman, R. Ragupathy,”Iris Recognition for Personal
Identification System”, ICMOC-2012, India
particular polling booth. The iris pattern of the person is
[12] Z.Zainal Abidin, M.Manaf, A.S.Shibghatullah, S.H.A.Mohd Yunos,
obviously unique. It reduces the polling time which is most S.Anawar, Z.Ayop,” Iris Segmentation Analysis using Integro-
important. It totally rules out the chance of invalid vote. Differential Operator and Hough Transform in Biometric
System”,Journal of Telecommunication, Electronic and Computer
Engineering, ISSN: 2180 - 1843 Vol. 4 No. 2 July - December 2012.
REFERENCES
[13] U T Tania, S M A Motakabber and M I Ibrahimy,” Edge Detection
[1] K.Ramya Devi and J.V.Vidhya,” SURVEY ON SECURE Techniques for Iris Recognition System”, IOP Conference Series:
ELECTRONIC VOTING SYSTEM”,International Journal of Materials Science and Engineering, Volume 53, conference 1, 2013.
Pharmacy & Technology (IJPT), Vol. 9,April-2017.
[14] K. Seetharaman, R. Ragupathy,” Iris Recognition for Personal
[2] Asif Ahmed Anik,Rayeesa Jameel, Abul Farah Anik, Nowroze Akter, Identification System”, Procedia Engineering, Volium 38 p:1531 –
"Design of a solar power Electronic Voting Machine", Proceedings of 1546, 2012.
2017 International Conference on Networking, Systems and Security
(NSysS),5-8 Jan. 2017,Dhaka, Bangladsh.
[3] Htet Ne Oo, Aye Moe Aung,"Design and formal analysis of
electronic voting protocol using AVISPA",Proceedings of 2017 2nd
International Conference for Convergence in Technology (I2CT), 7-9
April 2017, Mumbai, India.
[4] Supeno Djanali, Baskoro Adi Pratomo, Karsono Puguh Nindyo Cipto,
Astandro Koesriputranto, Hudan Studiawan, "Design and
development of voting data security for electronic voting (E-Voting)"
Proceedings of 2016 4th International Conference on Information and
Communication Technology (ICoICT), 25-27 May 2016, Bandung,
Indonesia.
[5] Zuyina Ayuning Saputri, Amang Sudarsono, Mike Yuliana, "E-voting
security system for the election of EEPIS BEM president", 2017
International Electronics Symposium on Knowledge Creation and
Intelligent Computing (IES-KCIC), 26-27 Sept. 2017, Surabaya,
Indonesia.
[6] Noureddine Cherabit, Fatma Zohra Chelali, Amar Djeradi,” Circular
Hough Transform for Iris localization” Science and Technology, p-
ISSN: 2163-2669, e-ISSN: 2163-2677, 2(5): 114-121, 2012;
[7] Krzysztof Misztal, Emil Saeed, Jacek Tabor, and Khalid Saeed,” Iris
Pattern Recognition with a New Mathematical Model to Its Rotation
Detection”, Human Identification by Vascular Pattern (pp.43-65),
November 2012.