Conference PaperPDF Available

Classification of Morphological Iris Properties Using Naïve Bayes Classifier

Authors:

Abstract and Figures

In recent years, person recognition and identity inquiry have been frequently encountered in many areas. Especially in places such as private office entrances, airports, banks, this type of person recognition is done under much more stringent controls. Biometric systems emerged with the idea that using human beings as a control tool would be the most reliable way. Biometric systems mean the use of human physiological features such as iris, fingerprint, face, hand for identification purposes. In this study, iris classification was performed on an iris database that can be used in biometrics. For the classification process, Naive Bayes classifier which is the machine learning algorithm was used. The result was 96.04% education and 92.11% test success.
Content may be subject to copyright.
IV Mühendislik, 07- 2019,
IV. International Scientific and Vocational Studies Congress - Engineering, 07-10 November 2019, /TURKEY
10
BILMES EN 2019 - ANKARA
CLASSIFICATION OF MORPHOLOGICAL IRIS PROPERTIES USING NAIVE BAYES
CLASSIFIER
Emre Avuçlu1, Abdullah Elen2
1Department of Computer Technology, Aksaray University, Aksaray/TURKEY
2Department of Computer Technology, Karabuk University, Karabuk/TURKEY
emreavuclu@aksaray.edu.tr, aelen@karabuk.edu.tr
Abstract
In recent years, person recognition and identity inquiry have been frequently encountered in many areas. Especially in places such as
private office entrances, airports, banks, this type of person recognition is done under much more stringent c ontrols. Biometric
systems emerged with the idea that using human beings as a control tool would be the most reliable way. Biometric systems mean the
use of human physiological features such as iris, fingerprint, face, hand for identification purposes. In this study, iris classification
was performed on an iris database that can be used in biometrics. For the classification process, Naive Bayes classifier which is the
machine learning algorithm was used. The result was 96.04% education and 92.11% test success.
Keywords: Biomedical diagnostics, Machine learning algorithms, Iris database.
1. Introduction
Biometric systems analyze certain parameters according to the method used. The number of these parameters depends on the
physical property to be scanned. It is a general technique that measures the physiological characteristics of biometric individuals and
is used in diagnostic and verification applications using these features [1]. Biometric systems, just like the human brain, can
recognize and differentiate. Instead of the identifiers that can be lost, stolen or forgotten, such as cards and passwords, the direct use
of the person makes the system more reliable and convenient. The devices used in biometric systems analyze the physical
characteristics that are different in each person and authenticate them for access to environments such as databases, banks and
computer systems without using a password.
Boles et al. used edge detection algorithms to locate the iris [2]. Sanchez-Reillo et al. found the iris margin information using
histogram analyzes. In the comparison stage, they applied the Hamming distance method [3]. Tisse et al. used the Hough
transformation to determine the location of the iris. They applied the two -dimensional Hilbert transform in the feature extraction
stage and the Hamming distance method in the comparison stage [4]. Lim et al. found the inner and outer edges of the iris using edge
detection algorithms. They performed 94.4% accurate classification in their study on 200 people [5]. Huang et al. used an integro-
differential operator for iris localization. In the comparison stage, they applied the Euclidean distance method. They achiev ed 100%
accurate recognition in their practice on 10 people [6]. Alim et al. applied Hough transformation to determine the location of the iris.
In their study of 159 people, they achieved recognition with an accurate classification rate of 96.1% [7]. Liam et al. determined the
inner and outer margins of the iris using edge detection algorithms. In their study, they used iris images from 30 individuals and
obtained 83% correct classification [8]. Yu et al. looked at the similarity of two iris information using fractional Fourier transform in
the comparison step [9]. Szewczyk et al. performed an iris recognition using the multi-layer perceptron neural network model in the
comparison phase [10]. Using a fast modular neural network, El-Bakry looked for iris in a picture [11]. Kong et al. proposed an
algorithm for detecting light reflections and lash noise in the i mage to improve the performance of iris localization [12]. Vatsa et al.
performed a study comparing 4 different iris recognition algorithms [13]. In this study, properties of 3 different iris images were
classified with Naive Bayes classifier by using iris database.
2. Materials and Methods
In this section, information about the database and Naive Bayes machine learning algorithm is given.
2.1. Dataset and its properties
The dataset contains three classes (Iris Setosa, Iris Versicolour, Iris Virginica) of 50 instances each, where each class refers to a type
of iris plant. One class is linearly separable from the other 2; the latter are not linearly separable from each other. Predicted attribute:
class of iris plant. This is an exceedingly simple domain. This data differs from the data presented in Fishers article. The 35th sample
should be: 4.9, 3.1, 1.5, 0.2, "Iris-setosa" where the error is in the fourth feature. The 38th sample: 4.9, 3.6, 1.4, 0.1, "Iris-setosa"
where the errors are in the second and third features [14].
1. sepal length in cm
2. sepal width in cm
3. petal length in cm
4. petal width in cm
IV Mühendislik, 07- 2019,
IV. International Scientific and Vocational Studies Congress - Engineering, 07-10 November 2019, /TURKEY
11
BILMES EN 2019 - ANKARA
5. class:
Iris Setosa
Iris Versicolour
Iris Virginica
2.2. Naive Bayes Classifier
The Naïve Bayes classifier is a simple probabilistic classification method based on Bayes theorem. In the Bayes theorem (BT), where
two independent events (and) occur one after the other, the probability of the second event occurring in the event of one of these two
events can be shown. By means of the change feature, the product rule as in Eq. 1 can be written with two different expressions;
P X Y P X Y P Y P Y X P X
(1)
Bayes' theorem defines the relationship between a random event that arises from a random process and conditional probabilities and
marginal probabilities for another random event as in Eq. 2.
P X Y P Y
P Y X P X (2)
The probabilities of the dependent states that are likely to occur in any problem are calculated by the Bayes equation given above. In
this equation, ( ) represents the input probability of the problem, ( ) represents the probability of a possible exit status, and (
| ) represents the probability of a Y output versus input [15].
3. Experimental Results
This section provides information about the performance of the system using Confusion Matrix. It is a matrix model that provides a
holistic approach to the classification performance of an intelligent system algorithm. A confusion matrix is structurally expressed as
in equation 3 below.
TP FP
CM FN TN
(3)
In this study, 10 statistical measurements were used to analyze the results. These measurements and formulas are shown in Table 1.
Table 1. Statistical measurement methods
Sensitivity or
True Positive Rate
TP
TPR TP FN
Dice Similarity
Coefficient
2
2
TP
DSC TP FP FN
Specificity or
True Negative Rate
TN
TNR TN FP
Accuracy
TP TN
ACC TP TN FP FN
Precision or
Positive Predictive Value
TP
PPV TP FP
Negative Predictive Value TN
NPV TN FN
F-Measurements
2
1 1
FM
TPR PPV
False Positive Ratio
FP
FPR TN FP
Matthews Correlation Coefficient
TP TN FP FN
MCC TP FP TP FN TN FP TN FN
False Negative Ratio
FN
FNR TP FN
The training and test ratios of the dataset used in the experimental studies were 73% and 27%, respectively. Table 2 below gives the
results of the statistical measurements obtained according to these diagnostic procedures.
IV Mühendislik, 07- 2019,
IV. International Scientific and Vocational Studies Congress - Engineering, 07-10 November 2019, /TURKEY
12
BILMES EN 2019 - ANKARA
Table 2. Statistical measurement results
TPR SPC PPV NPV FPR FNR ACC MCC FM
Class0: Iris Setosa 1,00 1,00 1,00 1,00 0,00 0,00 1,00 1,00 1,00
Class1: Iris Versicolour 0,92 0,92 0,85 0,96 0,08 0,04 0,92 0,82 0,88
Class2: Iris Virginica 0,86 0,96 0,92 0,92 0,04 0,08 0,92 0,83 0,89
The following figure shows the iris classification results with Naive Bayes algorithm.
Figure 1: Iris classification results with Naive Bayes algorithm.
4. Conclusions
In recent studies, we can see that iris images are classified using machine learning algorithms. In this study, Naive Bayes, one of the
machine learning algorithms, was used to classify the data in an iris dataset. It is aimed to identify the iris images in the dataset with
the verification rate. In this context, test results were evaluated by using statistical methods. In this study, data were prepared by
using four different morphological features. 92.11% test success was achieved with Naive Bayes classifier obtained from three
different iris classes.
5. References
[1]. Wildes, R.P. 1997. Iris Recognition: An Emerging Biometric Technology. Proceedings of the IEEE Vol. 85 No. 9.
[2]. Boles, W.W. and Boashah B. 1998. A human identification Technique Using Images of the Iris and Wavelet Transform,
IEEE Tr. on Signal Proces., Vol. 46, pp. 1185-1188.
[3]. Sanchez-Avila, C. and Sanchez-Reillo R. 2002. Multiscale Analysis for Iris Biometrics. Proc. of Int. Carnahan Conference
on Security Technology, pp 35-38.
[4]. Tisse, C.-L., Martin, L., Torres, L. and Robert, M. 2002. Person identification technique using human iris recognition. The
15th International Conference on Vision Interface, pp.294 299.
[5]. Lim, S., Lee, K., Byeon, O. and Kim, t. 2001. Efficient Iris Recognition through Improvement of Feature Vector and
Classifier, ETRI J., Vol. 23, No. 2, PP. 61- 70.
[6]. Huang, Y.P., Luo, S.W. and Chen, E.Y. 2002. An efficient iris recognition system. Proc.1st Int. Conf. Mach. Learning and
Cybermetics, Beijing.
[7]. Alim, O. ve Sharkas M. 2002. Texture classification of the human iris recognition using artificial neural Networks. Proc. of
IEEE Melecon 2002, Cairo.
[8]. Liam, L., Chekima, A., Fan, L. and Dargham, J. 2002. Iris recognition using selforganizing neural network. In IEEE, 2002
Student Conference on Research and Developing Systems, pp. 169 172, Malaysia.
[9]. Yu, L., Wang, K., Wang, C. And Zhang, D. 2002. Iris verification based on fractional fourier transform. Proc. of 1st Int.
Conf. on Machine Learning and Cybernetics, Beijing.
IV Mühendislik, 07- 2019,
IV. International Scientific and Vocational Studies Congress - Engineering, 07-10 November 2019, /TURKEY
13
BILMES EN 2019 - ANKARA
[10]. Szewczyk, R., Jablonski, P. Kulesza, Z., Napieralski, A., Cabestany, J. and Moreno, M. 2002. Automatic people
identification on the basis of the iris pattern extraction features and classification. Proc. of 23rd Int. Conf. On Microelectronics, vol.
2, Yugoslavia.
[11]. El-Bakry H. M.2001. Human iris detection using fast cooperative modular neural networks. Int. Joint Conf. on Neural
Networks, Washington, Dc., USA, 14-19 July, 577-582.
[12]. Kong, W.K. and Zhang, D. 2001. Accurate iris segmentation based on novel reflection and eyelash detection model, in
Proceedings of Int. Symp. Ov Intelligent Multimedia, Video and Speech Processing, Honk Kong, pp. 263-266.
[13]. Vatsa, M., Singh, R. and Gupta, P. 2004. Comparison of iris recognition algorithms. International Conference on Intelligent
Sensing and Information Processing India, pp.354-358.
[14]. Web site: https://archive.ics.uci.edu/ml/datasets/Iris, accessed date:19.9.2019.
[15]. -Elektronik ve Bilgisayar
zyumu, Bursa, 722-724 (2012).
ResearchGate has not been able to resolve any citations for this publication.
Conference Paper
Full-text available
In this paper, we have studied various well known algorithms for iris recognition. Four algorithms due to Sanchez-Avila et al. (2001), Li Ma et al. (2002), Tisse et al. (2002) and Daugman (2001) are implemented and compared on the CASIA iris image database. The results show that the Daugman's algorithm gave the highest accuracy of 99.9%.
Conference Paper
Full-text available
A combination of fast and cooperative modular neural nets to enhance the performance of the detection process is introduced. I have applied such an approach successfully to detect human faces in cluttered scenes (El-Bakry et al.) (2000). Here, this technique is used to identify human irises automatically in a given image. In the detection phase, neural nets are used to test whether a window of 20×20 pixels contains an iris or not. The major difficulty in the learning process comes from the large database required for iris/non-iris images. A simple design for cooperative modular neural nets is presented to solve this problem by dividing these data into three groups. Such division results in reduction of computational complexity and thus decreasing the time and memory needed during the test of an image. Simulation results for the proposed algorithm show a good performance. Furthermore, faster iris detection is obtained through image decomposition into many sub-images and applying cross correlation in the frequency domain between each sub-image and the weights of the hidden layer
Article
In this paper, we propose an efficient method for personal identification by analyzing iris patterns that have a high level of stability and distinctiveness. To improve the efficiency and accuracy of the proposed system, we present a new approach to making a feature vector compact and efficient by using wavelet transform, and two straightforward but efficient mechanisms for a competitive learning method such as a weight vector initialization and the winner selection. With all of these novel mechanisms, the experimental results showed that the proposed system could be used for personal identification in an efficient and effective manner.
Conference Paper
Iris verification is one of the biometrics verification technologies. This paper proposes a new iris verification method based on fractional Fourier transform. Through comparing two irises' fractional Fourier transform, we can distinguish the people whether they are the same person. At last, we introduce some applications of iris verification used for security in e-commerce.
Conference Paper
Iris recognition, a relatively new biometric technology, has great advantages, such as variability, stability and security, and is most promising for high security environments. A new iris recognition algorithm is proposed in this paper, which adopts Independent Component Analysis (ICA) to extract iris texture feature and a competitive learning mechanism to recognize iris patterns. Experimental results show that the algorithm is efficient and adaptive to the environment, e.g. it works well even for blurred iris images, variable illumination, and interference of eyelids and eyelashes.
Conference Paper
In this work, a new biometric identification approach based on the human iris pattern is proposed. The main idea of this technique is to represent features of the iris by multiscale analysis of the corresponding discrete dyadic wavelet transform zero-crossing representation. In this study we consider the iris signature as the mean of the gray values of different contours of virtual circles in a determined ring-shaped iris region. The resulting one-dimensional signals are compared with model features using different distances. Results show 98.7% classification success, achieving an equal error rate below 0.2% using Hamming distance, and the possibility of having false acceptance, rates with low false rejection rates.
Conference Paper
Among biometric systems for user verification, iris recognition systems represent a relatively new technology. Our system consists of two main parts: a localizing iris and iris pattern recognition. The raw image is captured using a digital camera. The iris is then extracted from the background after enhancement and noise elimination. Due to noise and the high degree of freedom in the iris pattern, only parts of the iris structure are selected for recognition. The selected iris structure is then reconstructed into a rectangle format. Using a trained self-organizing map neural network, iris patterns are recognized. The overall accuracy of our network is found to be about 83%.
Conference Paper
Automatic personal identification systems have assumed great importance in the past few years. The iris of the human eye has a texture that is unique for each individual and remains stable over the years. In this paper two feature extraction techniques that are based on 2D Gabor wavelets and 2D DCT are suggested and compared with each other. The features obtained are fed to neural network classifiers for identification. The achieved recognition rate using the DCT coefficients was about 96% compared to 92% obtained using the Gabor coefficients.
Conference Paper
In this paper an attempt to design decision part of the iris identification system, which will be able to identify persons just by a look at the camera, is discussed. Proposed system will be cheap in comparison with present ones. Modified Haralick's co-occurrence method with multilayer perceptron is used for extraction and classification of the irises. Some preliminary research results are also presented
Article
This paper examines automated iris recognition as a biometrically based technology for personal identification and verification. The motivation for this endeavor stems from the observation that the human iris provides a particularly interesting structure on which to base a technology for noninvasive biometric assessment. In particular the biomedical literature suggests that irises are as distinct as fingerprints or patterns of retinal blood vessels. Further, since the iris is an overt body, its appearance is amenable to remote examination with the aid of a machine vision system. The body of this paper details issues in the design and operation of such systems. For the sake of illustration, extant systems are described in some amount of detail