International Journal of Trend in Scientific Research and Development (IJTSRD)

Volume 3 Issue 5, August 2019 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470

Software Implementation of Iris

Recognition System using MATLAB
Mo Mo Myint Wai1, Nyan Phyo Aung2, Lwin Lwin Htay3
1Lecturer, 2Associate
Professor, 3Professor
1, 2, 3Department of Electronic Engineering, Technological University (Mandalay), Mandalay, Myanmar

How to cite this paper: Mo Mo Myint Wai ABSTRACT

| Nyan Phyo Aung | Lwin Lwin Htay The software implementation of iris recognition system introduces in this
"Software Implementation of Iris paper. This system intends to apply for high security required areas. The
Recognition System using MATLAB" demand on security is increasing greatly in these years and biometric
Published in recognition gradually becomes a hot field of research. Iris recognition is a
International branch of biometric recognition method. In thesis, Iris recognition system
Journal of Trend in consists of localization of the iris region and generation of data set of iris
Scientific Research images followed by iris pattern recognition. In thesis, a fast algorithm is
and Development proposed for the localization of the inner and outer boundaries of the iris
(ijtsrd), ISSN: 2456- region. Located iris is extracted from an eye image, and, after normalization
6470, Volume-3 | IJTSRD25258 and enhancement, it is represented by a data set. Using this data set a Neural
Issue-5, August Network (NN) is used for the classification of iris patterns. The adaptive
2019, pp.290-295, learning strategy is applied for training of the NN. The implementation of the
https://doi.org/10.31142/ijtsrd25258 system is developed with MATLAB. The results of simulations illustrate the
effectiveness of the neural system in personal identification. Finally, the
Copyright © 2019 by author(s) and accuracy of iris recognition system is tested and evaluated with different iris
International Journal of Trend in Scientific images.
Research and Development Journal. This
is an Open Access article distributed KEYWORDS: Iris, Biometric, Neural Network, MATLAB
under the terms of I. INTRODUCTION
the Creative Biometrics technology plays important role in public security and information
Commons Attribution security domains. Using various physiological characteristics of human, such as
License (CC BY 4.0) face, facial thermo grams, fingerprint, iris image, retinal scans and hand
(http://creativecommons.org/licenses/by geometry etc.,
/4.0)
biometrics accurately identify each individual and security and access control applications to mean measurable
distinguishes one from another . Iris recognition is one of physical characteristics of a person that can be checked on
important biometric recognition approach in a human an automated basis. Security personnel look for biometric
identification is becoming very active topic in research and data that does not change over the course of the life; that is,
practical application. Iris region is the part between the pupil they look for physical characteristics that stay constant and
and the white sclera. This field is sometimes called iris that are difficult to fake or change on purpose.
texture. The iris texture provides many minute
characteristics such as freckles, coronas, stripes, furrows, A. Biometric Technology
crypts, etc. These visible characteristics are unique for each A biometric system provides automatic recognition of an
subject. The human iris is not changeable and is stable. From individual based on some sort of unique feature or
one year of age until death, the patterns of the iris are characteristic possessed by the individual. Biometric
relatively constant over a person’s lifetime [1-4]. Because of systems have been developed based on fingerprints, facial
this uniqueness and stability, iris recognition is a reliable features, voice, hand geometry, handwriting, the retina [],
human identification technique. and the one presented in this thesis, the iris. Biometric
systems work by first capturing a sample of the feature, such
Iris recognition consists of the iris capturing, pre-processing as recording a digital sound signal for voice recognition, or
and recognition of the iris region in a digital eye image. Iris taking a digital color image for face recognition. The sample
image preprocessing includes iris localization, normalization, is then transformed using some sort of mathematical
and enhancement. In iris localization step, the determination function into a biometric template. The biometric template
of the inner and outer circles of the iris and the determination will provide a normalized, efficient and highly discriminating
of the upper and lower bound of the eyelids are performed. In representation of the feature, which can then be objectively
this research, the iris recognition system is developed using compared with other templates in order to determine
neutral network with MATLAB. identity. Most biometric systems allow two modes of
operation. An enrolment mode for adding templates to a
II. THE BACKGROUND THEORY OF BIOMETRIC database, and an identification mode, where a template is
TECHNOLOGY created for an individual and then a match is searched for in
It includes iris history and iris recognition system. The word the database of pre-enrolled templates.
biometric can be defined as "life - measure." It is used in

@ IJTSRD | Unique Paper ID – IJTSRD25258 | Volume – 3 | Issue – 5 | July - August 2019 Page 290
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
B. Different Types of Biometric Technologies degeneration occurs, resulting in the pupil opening and the
There are many types of biometric technologies as follow []. random, unique patterns of the iris. Although genetically

Body Odour identical, an individual’s irides are unique and structurally

DNA distinct, which allows for it to be used for recognition

Ear Shape purposes.

Facial Recognition

Finger Scanning Before recognition of the iris takes place, the iris is located

Hand Geometry using landmark features. These landmark features and the

Iris Recognition distinct shape of the iris allow for imaging, feature isolation,

Keystroke and extraction. Localization of the iris is an important step in

Retinal Scan iris recognition because, if done improperly, resultant noise

Personal Signature (e.g., eyelashes, reflections, pupils, and eyelids) in the image

Vein Pattern may lead to poor performance.

Voice Recognition

C. Iris Theory
The iris is the plainly visible, colored ring that surrounds the
pupil. It is a muscular structure that controls the amount of
light entering the eye, with intricate details that can be
measured, such as striations, pits, and furrows. The iris is not
to be confused with the retina, which lines the inside of the
back of the eye. The figure 1 shows human eye
characteristics. No two irises are alike. There is no detailed
correlation between the iris patterns of even identical twins, Figure2. Iris Structure
or the right and left eye of an individual. The amount of
information that can be measured in a single iris is much A. Typical Stage of Iris Recognition
greater than fingerprints, and the accuracy is greater than This system presents an introduction to divergent aspects of
DNA. pattern recognition. The operation of a pattern recognition
system is presented as a series of consecutive processing
III. IRIS RECOGNITION SYSTEM stages in figure 5. Pattern recognition is an information-
The iris is the plainly visible, colored ring that surrounds the reduction process: the assignment of visual or logical
pupil. It is a muscular structure that controls the amount of patterns to classes based on the features of these patterns
light entering the eye, with intricate details that can be and their relationships.
measured, such as striations, pits, and furrows. The iris is not
to be confused with the retina, which lines the inside of eye
of an individual. Fig 2.1 shows human eye characteristics. No
two irises are alike. There is no detailed correlation between
the iris patterns of even identical twins, or the right and left
eye of an individual. The amount of information that can be
measured in a single iris is much greater than fingerprints,
and the accuracy is greater than DNA.
Figure5. Block Diagram Iris Recognition Process [7]
Iris recognition is the process of recognizing a person by
analyzing the random pattern of the iris as shown Fig 1. The IV. THE PROPOSED IRIS RECOGNITION SYSTEM
automated method of iris recognition is relatively young, In this work, the proposed iris recognition system using
existing in patent only since 1994. The iris is a muscle within Artificial Neural Network is shown in Fig 6. The process for
the eye that regulates the size of the pupil, controlling the iris recognition system can be divided into the following
amount of light that enters the eye. It is the colored portion different stages:
of the eye with coloring based on the amount of melatonin
Image acquisition
pigment within the muscle as shown in Figure 2.
Edge detection

Localization

Feature extraction

Creating and training the neural networks

Figure1. Anatomy of the Eye [6]

Although the coloration and structure of the iris is

genetically linked, the details of the patterns are not. The iris
develops during prenatal growth through a process of tight Figure6. Block diagram of software implementation of iris
forming and folding of the tissue membrane. Prior to birth, recognition process

@ IJTSRD | Unique Paper ID – IJTSRD25258 | Volume – 3 | Issue – 5 | July - August 2019 Page 291
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
A. Image Acquisition B. Edge Detection
The iris recognition system includes iris image acquisition Canny edge operator is chosen to extract the clear edge of
and iris recognition. The iris image acquisition includes the iris and pupil area. Some other operators are Robert,
lighting system, the positioning system, and the physical Prewitt, and sobel. By analyzing the output edge images, the
capture system. In iris acquisition, the iris image in the input canny image is found that as the best operator for the pupil
sequence must be clear and sharp. Clarity of the iris’s minute detection as shown in Fig 9.
characteristics and sharpness of the boundary between the
pupil and the iris, and the boundary between the iris and the
sclera affects the quality of the iris image. A high quality
image must be selected for iris recognition. However, the eye
pictures are taken while trying to maintain appropriate
settings such as lighting and distance to camera, the light
spot from the camera cannot be removed from the image.
The preprocessing is difficult to detect the iris area from
those pictures. Therefore, CASIA iris [5] database is used in
this thesis. The CASIA [8] database used a special camera in
the infrared spectrum of light, not visible by the human eye.
In CASIA, each iris class is composed of seven samples taken
in two sessions, three in the first session and four in the
second. Images are 320x280 pixels gray scale taken by a
digital optical sensor designed by NLPR (National
Laboratory of Pattern Recognition Chinese Academy of
Sciences). There are 108 classes or irises in a total of 756 iris
images. The train data set is taken 5 kinds of irises from Figure9. Edge Detection Methods for Sample Eye (a)
CASIA database. Fig 7 shows the example of train data set Robert (b) Prewitt (c) Sobel (d) Canny
three iris images are collected in each type. The train data set
contains 3x5 = 15 images in train folder. After collection of C. Iris Localization
train set, the program has to be preprocessing. In the By utilizing the eye image, the boundary between the pupil
preprocessing stage, the images are transformed from RGB and the iris is detected after the position of the eye in the
to gray level and from unsigned integer eight bit or double given image is localized. After the center and the radius of
precision thus facilitating the manipulation of the images in the pupil are extracted, the right and the left radius of the iris
subsequent steps in Fig 8. are searched based on these data. For this part, the following
sequences are conducted.

D. Feature Extraction
In order to provide the compact features for accurate
recognition of individuals, the most discriminating
information present in an iris pattern must be extracted.
Only the significant features of the iris must be encoded so
that the neural network can be classify well.

The feature is extracted by limiting 6x6 matrixes which is

divided the iris pattern for 6 rows and 6 columns. The
summation of the pixel values for each region is calculated
and recorded for each iris image. They are saved in train
feature vector set. The selected vectors with is checked by
the following table. Fig 10 is the flowchart of feature
extraction system.

E. Neural Network Creation and Training Process

After the features are saved in the file, the neural network is
created and trained with supervised learning. The target
vector is initialized with the relevant 1, 0 matrix. Logic 1 is
set for the recognized person position, and logic 0 is set for
other places. In the training part, goal is set to 0.01 of mean
square error, and number of epoch is set to 1000. After
Figure7. Example of Iris Train Data Set training process, network can be used for recognition parts.
The neural network training function is to apply according to
the below MATLAB code and the Fig 11.of flow chart.

F. Graphical User Interface (GUI)

GUIDE, the MATLAB Graphical User Interface development
environment, provides a set of tools for creating graphical
user interface (GUIs). These tools greatly simplify the
process of designing and building GUIs. The GUIDE tools can
be used to lay out the GUI. Using the GUIDE Layout Editor, a
Figure8. RGB Image and Gray Scale Images

@ IJTSRD | Unique Paper ID – IJTSRD25258 | Volume – 3 | Issue – 5 | July - August 2019 Page 292
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
GUI can be laid easily by clicking and dragging GUI
components, such as panels, buttons, text fields, sliders,
menus, and so on into the layout area.

V. SIMULATION RESULTS
MATLAB programming is used for the development and
simulation of iiris recognition which has many stages such as
iris recognition, localization is an important step to isolate
the iris region. Edge detection method has been described to
remove the pupil region from eye images. After localization,
feature extraction method and adaptive histogram
equalization method have been described. Then, the
selective feature vectors is used for classification of neural
network and the GUI software implementation is provided
for an identify person.

After the development of iris recongin system, the

performance of the proposal system and some tests are
carried out. The results for the system are explained in
detail. The proposed system for the identification includes
two operations modes: training process and identification
process. These processes are called recognition system using
GUI software implementation.

Figure11. Flow Chart for Neural Network Training and

Identified Program

GUIDE stores a GUI in two files, where are generated the first
time save or run the GUI and shows the creation of user
interface in Fig 12: The proposed iris recongitnion system is
implement the GUI based system using matlab as shown in
figure 12 .

Figure 12.Graphical User Interface of Software

Figure10. Flow Chart for Feature Extraction Program Development

@ IJTSRD | Unique Paper ID – IJTSRD25258 | Volume – 3 | Issue – 5 | July - August 2019 Page 293
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
In the first mode, training process is applied for the feature The adaptive histogram equalization is a useful method iris
vectors, histogram analysis and training neural network. The image processing for contrast adjustment using the image’s
isolating of iris is collected by feature vectors. Each of histogram.
feature vectors displays a histogram for the iris image with a
gray scale colorbar. The histogram of a gray scale image uses In order to evaluate the iris recognition algorithms the
a default value of 256 bins. CASIA iris image database is used. Currently this is largest
iris database available in the public domain. This image
If a grays scale image is a binary image, histogram uses two database contains 756 eye images from 108 different
bins. Then these vectors become the training data set for the persons. In this thesis the performance of the iris recognition
neural network. The training of recognition system is carried system as a whole is examined. Tests were carried out to find
out using grayscale value of iris images. Neural network is the best separations that the false match and false accept
trained with the selected iris images. The selected iris input rate is minimized, and to confirm that iris recognition can
layers obtain good classification results. The output layer perform accurately as a biometric for recognition of
will be continued as much neuron as there are class to individuals. As well as, confirming that the system provides
recognize. In this process, the user has to extract the accurate recognition, experiments were also conducted in
features. When the feature extraction part is finished, the order to confirm the uniqueness of human iris patterns by
output can be seen as in Fig 13. The mean square error is deducing the number of degree of freedom present in the iris
decreasing with the epoch. When the goal is met or the template representation.
predefined epoch is arrived, the training process will be
stopped. The iris recognition can be test with GUI interface as shown
in figure 14.

Figure13. Feature Database Display

The open file contains the trained eye image from CASIA
database. The gray scale eye images were acquired from Figure14. Graphical User Interface of Iris Recognition
database. It is choose one iris image for the identify person.
The edge detection is described differentiation form the Table 5.1 demonstrates the comparative results of different
foundation for many applications in computer vision. It iris images used for identification. In this process, only five
detects each edge point of selected eye image by canny train databases are applied for using Neural Network. The
operator. iris pattern classification using NN is performed. For each
type of iris database, the test data set consists of three from
The localization is the detection of the iris area between the trained data and four from the untrained data. Five
pupils and sclera because of the eye image contains not only person’s irises are selected from iris database for
the iris region but also some unuseful parts; such as the classification. The average time for the recognition of the iris
pupil, eyelids, and sclera and so on. So the subparts of the images was 0.02 s for NN classification. So, the accuracies of
input data image is evaluated by separated process using correct classification are 100 % and 50 % for database 001,
localization. 002, 004 and 003, 005. The identification result obtained
using the neural network approach the success of its efficient
Feature extraction achieves the texture of the iris after use iris recognition.
pupils were extracted from the image.

Table 5.1 Performance Comparison of Accrual and Test Result

Iris No: of Correct for Accuracy No: of Correct for % Correct
database train file train file time for (NN) test file test file for test file
001 3 3 0.02 s 2 2 100 %

002 3 3 0.02 s 2 2 100 %

003 3 3 0.02 s 2 1 50 %

004 3 3 0.02 s 2 2 100 %

005 3 3 0.02 s 2 1 50 %

@ IJTSRD | Unique Paper ID – IJTSRD25258 | Volume – 3 | Issue – 5 | July - August 2019 Page 294
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
VI. CONCLUSION method can be used in security purposes because Iris
Analysis of the developed iris recognition system has pattern is different from one person to another so that this
revealed a number of interesting conclusions. It can be can be highly reliable more than security system.
stated that localization is the critical stage of iris recognition,
since areas that are wrongly identified as iris regions will REFERENCES
corrupt biometric templates resulting in very poor [1] J. Daugman, Biometric Personal Identification System
recognition. The results have also shown that localization Based on Iris Analysis, United States Patent, no.
can be the most difficult stage of iris recognition because its 5291560, 1994.
success is dependent on the imaging quality of eye images.
[2] J. Daugman, “Statistical Richness of Visual Phase
So, these input images have been used the off line data using
Information: Update on Recognizing Persons by Iris
database from CASIA according to image acquisition steps.
Patterns,” Int’l J. Computer Vision, vol. 45, no.1, pp. 25-
The iris database provides images that are pixels.
38 , 2001
Another interesting finding was that the texture extraction
for the iris area. The localized iris images have been isolated [3] J. Daugman, “Demodulation by Complex-Valued
after cropping the pupil in the original image with threshold Wavelets for Stochastic Pattern Recognition,” Int J.
value. The result of feature extraction from resized localized Wavelets, Multi resolution and Information Processing,
image is captured the function of feature_vector. Every vol.1, no.1, pp.1-17, 2003.
Test_feature were segmented into matrix. The feature [4] J. Daugman, “How Iris Recognition Works”, University
vectors btained from feature extraction have been used as of Cambridge, 2001.
inputs to neural networks for classification, trained with 107
epochs. The well trained neural network can enhance the [5] https://www.medicalnewstoday.com/articles/320608
performance of the system. The classification of trained .php
network is identified the iris patterns. The recognition

@ IJTSRD | Unique Paper ID – IJTSRD25258 | Volume – 3 | Issue – 5 | July - August 2019 Page 295