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

Volume 7 Issue 5, September-October 2023 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470

Performance of Hasty and Consistent Multi

Spectral Iris Segmentation using Deep Learning
Ram Niwas Sharma1, Ankit Kumar Navalakha2, Neha Sharma3
1
M Tech Scholar, Department of Computer Science Engineering, Mewar University, Chittorgarh, Rajasthan, India
2
Assistant Professor, Department of Computer Science Engineering, Mewar University, Chittorgarh, Rajasthan, India
3
Lecturer, Department of Computer Science Engineering, MLVTE, Bhilwara, Rajasthan, India

ABSTRACT How to cite this paper: Ram Niwas

The recognition system is composed of seven phases: acquisition, Sharma | Ankit Kumar Navalakha | Neha
preprocessing, segmentation, normalization, feature extraction, Sharma "Performance of Hasty and
feature selection, and classification. In the acquisition phase, iris Consistent Multi Spectral Iris
images are captured, followed by preprocessing to enhance the Segmentation using Deep Learning"
Published in
quality of the images. The segmentation phase involves separating
International Journal
the iris region from the background, and the normalized iris region is of Trend in
shaped into a rectangle in the normalization phase. Iris segmentation Scientific Research
is a critical step in iris recognition systems and has a direct impact on and Development
authentication and recognition results. However, standard (ijtsrd), ISSN: 2456-
segmentation techniques may not perform well in noisy iris databases 6470, Volume-7 | IJTSRD59853
captured under challenging conditions. Moreover, the lack of large Issue-5, October
iris databases hinders the performance improvement of convolution 2023, pp.11-15, URL:
neural networks. The proposed method addresses these challenges by www.ijtsrd.com/papers/ijtsrd59853.pdf
effectively handling irregular iris images captured under visible light.
Copyright © 2023 by author (s) and
The iris region is processed and evaluated to generate a unique International Journal of Trend in
feature vector, which is then used for person identification. VGG16, a Scientific Research and Development
well-known deep learning model, is employed for image Journal. This is an
classification, and the feature vector is fed into VGG16 for Open Access article
classification purposes. distributed under the
terms of the Creative Commons
KEYWORDS: Deep Learning, Multi Spectral Iris, neural networks, Attribution License (CC BY 4.0)
VGG16 (http://creativecommons.org/licenses/by/4.0)

INTRODUCTION
A biometric system is an automated system that externally visible and protected organ, maintaining its
recognizes individuals based on unique features or stable epigenetic pattern throughout adulthood. The
characteristics. Various biometric systems have been visual system plays a crucial role in processing
developed using different features such as information within the human body [2]. An integral
fingerprints, facial features, voice, hand geometry, component of the visual system is the eye, which
handwriting, the retina, and the focus of this thesis, consists of three concentric layers. Biometric research
the iris [1]. Biometric systems typically operate in and development require the analysis of human data.
two modes: an enrollment mode for adding templates However, it is impractical to perform algorithm
to a database and an identification mode where a testing on real-time captured data due to various
template is created for an individual and matched constraints [3]. Therefore, standardized biometric
against the pre-enrolled templates in the database. . databases play a crucial role in the testing and
The iris comprises several layers, including the evaluation of recognition methods.
epithelium layer with densely pigmented cells and the
Biometrics encompasses quantifiable data related to
stromal layer with blood vessels, pigment cells, and
human characteristics and traits. It is used for
iris muscles [1]. The color of the iris is determined by identification and access control in computer science
the density of stromal pigmentation.
and surveillance applications. In the initial phase of
The epigenetic nature of iris patterns results in each biometric processing, various characteristics are
eye of an individual having independent patterns, captured. However, for automated capturing and
even in the case of identical twins. The iris is an

@ IJTSRD | Unique Paper ID – IJTSRD59853 | Volume – 7 | Issue – 5 | Sep-Oct 2023 Page 11

 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
comparison with stored data, the biometric Viktor Varkarakis and Shabab Bazrafkan (2018)
characteristics should satisfy certain criteria. developed an effective deep learning method for iris
biometric authentication. The proposed system
Multimodal biometric systems utilize more than one
physiological or behavioral characteristic for utilizes semantic segmentation technology based on
enrollment, verification, or identification. These the UnEAT model to locate and extract the ROI
(Region of Interest) of the iris in an eye image.
systems are employed in applications such as border
entry/exit, access control, civil identification, and Mahmut Karakaya (2018) conducted a study on
network security to reduce false match and false non- biometric authentication using an effective low-
match rates, provide alternative means of enrollment complexity YOLOv3 tiny-based deep learning
or identification, and counter fraudulent attempts [2]. inference network.
The selection of a specific biometric for an
METHODOLOGY
application requires weighing these factors, as no The presence of noise regions in captured images
single biometric can meet all requirements for every increases the need for robust and adaptable
situation. segmentation techniques [11]. Various iris
LITERATURE REVIEW segmentation proposals have been described,
The concept of iris recognition was first proposed by including statistical approaches, alpha trimming
Dr Frank Burch, an ophthalmologist, in 1936. In operation, and checking the gray level transition.
1987, ophthalmologists Aran Safir and Leonard Flom Image segmentation refers to the division of an image
patented the idea, and they enlisted John Daugman in into multiple components. It is a critical step in
1989 to develop practical algorithms for iris
automated image processing systems as it forms the
recognition [12]. foundation for subsequent operations such as
Cheng-Shun Hsiao, Chih-Peng Fan, and Yin- description or recognition [4].
Tsung Hwang (2022) proposed an algorithm for
VGG16 is a convolutional neural network (CNN)
locating the pupil center using an edge detection model proposed by K. Simonyan and A. Zisserman
method. They recorded the grey level values on
from the University of Oxford in their paper titled
virtual concentric circles and constructed a zero- "Very Deep Convolution Networks for Large-Scale
crossing representation based on a one-dimensional Image Recognition." This model gained popularity
dyadic wavelet transform. for its remarkable performance in the ImageNet
dataset, achieving a top-5 test accuracy of 92.7%.

Figure 01: VGG16 Architecture

The VGGNet configurations are labeled as A, B, C, D, and E. All configurations share the same overall
architecture and differ primarily in terms of depth. In terms of the width of the convolution layers [5], the
number of channels starts at 64 in the first layer. After each max-pooling layer, the number of channels doubles,
progressively increasing until it reaches 512 in the deeper layers [10].

@ IJTSRD | Unique Paper ID – IJTSRD59853 | Volume – 7 | Issue – 5 | Sep-Oct 2023 Page 12

 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
Software Description
MATLAB 19 (b) is a high-level technical computing language and interactive environment that offers various
capabilities such as algorithm development, data visualization, data analysis, and numerical computation [6]. It
provides strong support for working with matrices and matrix operations, making it a powerful tool for data
analysis and visualization.
One of the key features of MATLAB 19 (b) is its extensive collection of toolboxes, which are sets of programs
designed to support specific tasks. The image processing toolbox is particularly relevant for handling images and
offers functions, commands, and techniques tailored to image analysis [6]. In MATLAB, functions play a crucial
role and are used to perform specific tasks by accepting input parameters and producing output results.
Additionally, users have the flexibility to create their own functions when necessary. MATLAB treats matrices
as its standard data type, considering all data to be matrices in some form. Images, while strings are essentially
matrices of characters, with the string length determining the matrix dimensions.
EXPERIMENTAL RESULTS
The performance of various CNN architectures, including VGG16 and SVM models, was evaluated in this
study. The models were initialized with zero biases and random weights with zero mean [7]. Dropout was
applied to the fully connected layers to mitigate over fitting. By considering these performance parameters, the
classifier's accuracy, precision, sensitivity, and specificity can be evaluated, providing insights into its
classification capabilities.
Table 01: Comparison Result with Existing Work
Database Approach Accuracy (%)
EXISTING SVM 74
CASIA-INTERVAL
PROPOSED VGG16 90.91

Figure 02: Comparison Result with Existing Work

This implies that the VGG16 model demonstrated the highest classification accuracy among the tested
architectures. Additionally, Table 02 presents a comparison of different factors derived from various datasets,
providing further insights into the evaluation and comparison process [7].
Table 02 Comparison Result with Different Dataset
Accuracy Precision Recall F1 Specificity
Approach
(%) (%) (%) (%) (%)
CASIA-INTERVAL 90.91 91.44 91.0 89.49 91.38
UBRIS.V2 84.25 83.99 86.95 86.15 84.39
VGG16
MMU Database 87.15 83.26 84.12 85.15 86.00

@ IJTSRD | Unique Paper ID – IJTSRD59853 | Volume – 7 | Issue – 5 | Sep-Oct 2023 Page 13

 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470

Figure 03: Performance Graph

The evaluation was performed on the CASIA-INTERVAL dataset and compared with the UBRIS.V2 and MMU
databases [9]. The results indicated that CASIA-INTERVAL achieved the highest accuracy and specificity.
Furthermore, an attempt was made to utilize the conventional VGG16 architecture with pre-trained weights
obtained from a model in ImageNet [8].
CONCLUSION REFERNCES
The proposed system utilizes the GLCM and VGG16 [1] Caiyong Wang; Jawad Muhammad; Yunlong
for iris recognition. By combining these features with Wang; Zhaofeng; Zhenan Sun, "Towards
other texture generation processes, the feature Complete and Accurate Iris Segmentation
extraction process becomes more effective. The Using Deep Multi-Task Attention Network for
recognition of iris is achieved using the kernel Non-Cooperative Iris Recognition" IEEE
function of the VGG16 classifier. The system Transactions on Information Forensics and
accurately recognizes iris features in video, even in Security, 2022.
the presence of challenges such as illumination and [2] Cheng-Shun Hsiao; Chih-Peng Fan; Yin-Tsung
contrast variations. It also detects the liveliness and Hwang, "Design and Analysis of Deep-
race of the person with high accuracy. Extensive Learning Based Iris Recognition Technologies
experiments are conducted using the CASIA Iris-, by Combination of U-Net and EfficientNet"9th
UBIRIS.v2, and MMU databases to evaluate the International Conference on Information and
performance of the iris and eye recognition system Education Technology (ICIET), 2022.
using VGG16 architecture. The proposed system
outperforms other existing technologies on each [3] Chia-Wei Chuang; Chih-Peng Fan; Robert
database. Additionally, an iris segmentation Chen-Hao Chang, "Design of Low-Complexity
architecture based on CNN combined with VGG16 YOLOv3-Based Deep-Learning Networks with
and VGG19 is proposed to further improve the Joint Iris and Sclera Messages for Biometric
accuracy of the recognition system. Recognition Application", IEEE 9th Global
Conference on Consumer Electronics (GCCE),
FUTURE WORK 2022.
The Biometric Authentication can be improved by
including more number of possible features and other [4] Shabab; Peter Corcoran, "Enhancing iris
valid measures. Pupil dilation is found to have an authentication on handheld devices using deep
effect on the accuracy of iris recognition particularly learning derived segmentation techniques",
if the number of dilation is completely different at IEEE International Conference on Consumer
enrollment than at verification. Wearing contact Electronics (ICCE),2022.
lenses also modify the color and look of the attention [5] Mousumi Sardar; Subhash Banerjee; Sushmita
also can decrease the recognition rates. Mitra, "Iris Segmentation Using Interactive
Deep Learning", IEEE Access, 2021.

@ IJTSRD | Unique Paper ID – IJTSRD59853 | Volume – 7 | Issue – 5 | Sep-Oct 2023 Page 14

 International Journal of Trend in Scientific Research and Development @ www.ijtsrd.com eISSN: 2456-6470
[6] Gürkan Şahin; Orkun Susuz, "Encoder-Decoder Unconstrained Environments"31st SIBGRAPI
Convolutional Neural Network Based Iris- Conference on Graphics, Patterns and Images
Sclera Segmentation", 27th Signal Processing (SIBGRAPI), 2018.
and Communications Applications Conference [10] Lozej; Dejan, Vitomir, Peter Peer, "Influence of
(SIU), 2021.
segmentation on deep iris recognition
[7] Ehsan eddin Jalilian; Mahmut Karakaya; performance"7th International Workshop on
Andreas, "Iris Recognition Using CNN Based Biometrics and Forensics (IWBF), 2020.
Iris Segmentation", International Conference of
[11] Caiyong Wang; Yunlong Wang, Boqiang Xu;
the Biometrics Special Interest Group
Yong He, Zhiwei Dong; Zhenan Sun, "A
(BIOSIG), 2021.
Lightweight Multi-Label Segmentation
[8] Ananya Zabin, Thiri machos Bourlai, "A Deep Network for Mobile Iris Biometrics" IEEE
Learning Based Approach to Iris Sensor International Conference on Acoustics, Speech
Identification", IEEE/ACM International and Signal Processing (ICASSP), 2020.
Conference on Advances in Social Networks
[12] Daniel Kerrigan, Mateusz Trokie lewicz,
Analysis and Mining (ASONAM), 2021. Adam, Kevin W. Bowyer, "Iris Recognition
[9] Luiz A. Zanlorensi, Eduardo Luz, Rayson with Image Segmentation Employing Retrained
Laroca, Alceu S. Britto, Luiz S. Oliveira, David Off-the-Shelf Deep Neural Networks",
Menotti, "The Impact of Preprocessing on Deep International Conference on Biometrics (ICB),
Representations for Iris Recognition on 2020.

@ IJTSRD | Unique Paper ID – IJTSRD59853 | Volume – 7 | Issue – 5 | Sep-Oct 2023 Page 15