A NOVEL TECHNIQUE FOR IRIS RECOGNITION SYSTEM

Kamal Vahdati Bana

Day System Evaluation and Research Institute

Amin Rezaeian Delui

Toos Institute of Higher Education, Mashhad, Iran

Amir Azizi

Eqbal Institute of Higher Education, Mashhad, Iran

Keywords: Biometric, Iris Recognition, Contourlet Transform, Co-occurrence matrix.

Abstract: In this paper we propose a new feature extraction method for iris recognition based on contourlet transform.

Contourlet transform captures the intrinsic geometrical structures of iris image. It decomposes the iris image

into a set of directional sub-bands with texture details captured in different orientations at various scales so

for reducing the feature vector dimensions we use the method for extract only significant bit and

information from normalized iris images. In this method we ignore fragile bits. At last, the feature vector is

created by using Co-occurrence matrix properties. For analyzing the desired performance of our proposed

method, we use the CASIA dataset, which is comprised of 108 classes with 7 images in each class and each

class represented a person. And finally we use SVM and KNN classifier for approximating the amount of

people identification in our proposed system. Experimental results show that the proposed increase the

classification accuracy and also the iris feature vector length is much smaller versus the other methods.

1 INTRODUCTION

The purpose of ‘Iris Recognition’, a biometrical

based technology for personal identification and

verification, is to recognize a person from his/her iris

prints. In fact, iris patterns are characterized by high

level of stability and distinctiveness. Various iris

recognition methods have been proposed for

automatic personal identification and verification. In

Figure.2 you can see the typical stages of Iris

Recognition system. Daugman first presented a

prototype system (

Daugman, J., 1993, 2004) for iris

recognition based on multi-scale Gabor wavelets.

Wildes presented another iris recognition system

(

Wildes, R. P, 1996, et al) in which the iris pattern was

decomposed into multi-resolution pyramid layers

using wavelet transform. Both systems of Daugman

and Wildes employed carefully designed image

acquisition devices to get equal high quality iris

images. Zhenan presented a shift-invariant method

(

Zhenan Sun et al., 2005) which decomposed the iris

pattern into multiple bands using a two-dimensional

Gabor filter. Boles & B.Boashash (

Boles W.W and

Boashash B, 1998

) decomposed one-dimensional

intensity signals computed on circles in the iris and

use zero-crossings of the decomposed signals for the

feature representation. Section 2 deals with Iris

Recognition System overview. Section 3 deals with

Experimental results and discussion. Section 4

concludes this paper.

2 IRIS RECOGNITION SYSTEM

OVERVIEW AND PROPOSED

METHOD

Figure 1 illustrates the main steps of our proposed

Approach. First the image preprocessing step

performs the localization of the pupil, detects the iris

530

Vahdati Bana K., Rezaeian Delui A. and Azizi A. (2009).

A NOVEL TECHNIQUE FOR IRIS RECOGNITION SYSTEM.

In Proceedings of the International Joint Conference on Computational Intelligence, pages 530-533

DOI: 10.5220/0002282805300533

 SciTePress

Figure 1: Flow diagram of the proposed iris recognition scheme.

boundary, and isolates the collarette region, which is

regarded as one of the most important areas of the

iris complex pattern. The collarette region is less

sensitive to the pupil dilation and usually unaffected

by the eyelids and the eyelashes we also detect the

eyelids and the eyelashes, which are the main

sources of the possible occlusion. In order to achieve

the invariance to the translation and the scale, the

isolated annular collarette area is transformed to a

rectangular block of fixed dimension. The

discriminating features are extracted from the

transformed image and the extracted features are

used to train the classifiers. The optimal features

subset is selected using several methods to increase

the matching accuracy based on the recognition

performance of the classifiers.

2.1 Feature Extraction and Encoding

Only the significant features of the iris must be

encoded so that comparisons between templates can

be made. Gabor filter and wavelet are the well-

known techniques in texture analysis (

Ma, L, et al.,

2002, 2003)

(Daugman, J., 1993, 2004)

(

Zhu, Y. et al., 2000) In wavelet family, Haar wavelet

(

Jafar M. H. Ali, Aboul Ella Hussanien, 2003) was

applied by Jafer Ali to iris image and they extracted

an 87-length binary feature vector. The major

Drawback of wavelets in two-dimensions is their

limited ability in capturing directional information.

The contourlet transform is a new extension of the

wavelet transform in two dimensions using multi-

scale and directional filter banks. The feature

representation should have enough information to

classify various irises and be less sensitive to noises.

Also in the most appropriate feature extraction we

attempt to extract only significant information,

moreover reducing feature vector dimensions.

Therefore the processing lessened and enough

information is supplied to introduce iris feature

vectors classification.

2.1.1 Contourlet Transform

Contourlet transform (CT) allows for different and

flexible number of directions at each scale. CT is

constructed by combining two distinct

decomposition stages (

Do M. N., and Vetterli, M,

2004)

, a multi-scale decomposition followed by

directional decomposition. The grouping of wavelet

coefficients suggests that one can obtain a sparse

image expansion by applying a multi-scale

transform followed by a local directional transform.

It gathers the nearby basis functions at the same

scale into linear structures. In essence, a wavelet-like

transform is used for edge (points) detection, and

then a local directional transform for contour

segments detection. A double filter bank structure is

used in CT in which the Laplacian pyramid (LP)

(

Burt P. J and Adelson E. H, 1983) is used to capture

the point discontinuities, and a directional filter bank

(DFB) (

Bamberger R.H and Smith M. J. T, 1992) to link

point discontinuities into linear structures.

2.1.2 The Best Bits in an Iris Code

Iris biometric systems apply filters to iris images to

extract information about iris texture. Daugman’s

approach maps the filter output to a binary iris code.

Iris Code

Iris Image Pre Processing

Segmentation Normalization

Classification

11011011010

00101010101

11100011100

Feature

Comparison

Similarity

Value

Feature Extraction

A NOVEL TECHNIQUE FOR IRIS RECOGNITION SYSTEM

531

The fractional Hamming distance between two iris

codes is computed and decisions about the identity

of a person are based on the computed distance. The

fractional Hamming distance weights all bits in an

iris code equally. However, not all the bits in an iris

code are equally useful. For a given iris image, a bit

in its corresponding iris code is defined as “fragile”

if there is any substantial probability of it ending up

a 0 for some images of the iris and a 1 for other

images of the same iris.According to (

Karen P.

Hollingsworth, Kevin W. Bowyer, 2008) the percentage

of fragile bits in each row of the iris code, Rows in

the middle of the iris code (rows 5 through 12) are

the most consistent.

2.1.3 Feature Vector in Proposed Method

In our method we use the Grey Level Co-occurrence

Matrix (GLCM). The technique uses the GLCM

(Grey Level Co-occurrence Matrix) of an image and

it provides a simple approach to capture the spatial

relationship between two points in a texture pattern.

It is calculated from the normalized iris image using

pixels as primary information. Various textural

features have been defined based on the work done

by Haralick [Haralick, R.M, et al., 1973]. These

features are derived by weighting each of the co-

occurrence matrix values and then summing these

weighted values to form the feature value. The

specific features considered in this research are

defined as follows:

1) Energy

∑∑

jip

),(

2) Contrast

∑∑∑

−

===

⎥

⎦

⎤

⎢

⎣

⎡

=−

011

),(

ggg

njijiPn

3) Correlation =

jiPij

σσ

μμ

∑∑

−),()(

4) Homogeneity=

∑∑

−+

jiP

),(

)(1

5) Autocorrelation =

∑∑

jiPij ),()(

6) Dissimilarity =

∑∑

−

jiPji ),(.

7) Inertia =

∑

−

jiPji ),()(

Here

yxyx

,,,

are mean and standard

deviation along x and y axis.

3 EXPERIMENTAL RESULTS

For creating iris feature vector we carried out the

following steps:

1) Iris normalized image (Rows in the middle

of the iris code (rows 5 through 12)) is

decomposed up to level two.(for each

image ,at level one , 2 and at level two , 4

sub band are created ) .

2) The sub bands of each level are put

together, therefore at level one a matrix

with 4*120 elements, and at level two a

matrix with 16*120 elements is created. We

named these matrixes: Matrix1 and Matrix

3) By putting together Matrix1 and Matrix 2, a

new matrix named Matrix3 with 20*120

elements is created. The co-occurrence of

these three matrixes with offset one pixel

and angles 0, 45, 90 degree is created and

name this matrix: CO1, CO2 and CO3.in

this case for each image 3 co-occurrence

matrixes with 8*8 dimensions are created.

4) According to the Haralick‘s theory the co-

occurrence matrix has 14 properties , of

which in iris biometric system we used 7

properties which are used for 3 matrixes ,

so the feature vector is as follow:

F=[En1,Cont1,cor1,hom1,Acor1,dis1,ine1,

En2,Cont2,cor2, hom2,Acor2,dis2,ine2

En3,Cont3,cor3,hom3,Acor3,dis3,ine3]

In other word the feature vector in our method has

only 21 elements. Also for improving results, for

each sub bands and scale we create a feature vector

by using GLCM.in other words for each eight sub

bands in level 3 of Contourlet transform we

computed GLCM properties, separately and then by

combining these properties the feature vector is

created. In this case the feature vector has 56

elements. In Table 1 you can see the result of

implementing our proposed method:

IJCCI 2009 - International Joint Conference on Computational Intelligence

532

Table 1: Result of Implementing Proposed Method.

The

Number Of

Classes

The Correct of Percentage Classification (%)

KNN

Classifier

SVM

Classifier(Kernel

SVM

Classifier(Kernel

20 96.6 100 100

40 88.3 94.3 96.3

60 90.8 91.6 95.6

80 89.3 90.1 95.8

100(GLCM) 88.5 90.07 94.2

100(GLCM

(Combining

Sub bands)

87.5 91.3 96.3

4 CONCLUSIONS

In this paper we proposed an effective algorithm for

iris feature extraction using contourlet transform Co-

occurrence Matrix have been presented. The GLCM

proved to be a good technique as it provides

reasonable accuracy and is invariant to iris rotation.

For Segmentation and normalization we use

Daugman methods .Feature extraction in our

proposed method includes: sub bands proper

composition from Contuorlet pyramid and co-

occurrence calculations and finally selecting a set of

Haralick‘s properties that form the Maximum

distance between inter classes and Minimum

distance between intra classes. Our proposed method

can classify iris feature vector properly. The rate of

expected classification for the fairly large number of

experimental date in this paper verifies this claim. In

the other words our method provides a less feature

vector length with an insignificant reduction of the

percentage of correct classification.

REFERENCES

Bamberger, R.H., and Smith M. J. T, 1992. A filter bank

for the directional decomposition of images: Theory

and design, IEEE Trans. Signal Proc., vol. 40, no. 4,

pp. 882–893.

Boles, W.W., and Boashash B,1998.A Human

Identification Technique Using Images of the Iris and

Wavelet Transform, IEEE Trans. Signal Processing,

Vol.46, No.4, 1998, pp.1185-1188.

Burt P. J., and Adelson E. H., 1983. The Laplacian

pyramid as a compact image code, IEEE Trans.

Commun., vol. 31, no. 4, pp. 532–540.

Daugman, J., 1993. High confidence visual recognition of

persons by a test of statistical independence, IEEE

Transactions on Pattern Analysis and Machine

Intelligence, Vol. 15, No. 11, pp: 1148-1160.

Daugman, J., 2004. How Iris Recognition works.IEEE

Transactions on Circuits and systems for video

Technology, Vol.14, No.1, pp: 21-30.

Do M. N., and Vetterli, M, 2004. The contourlet

transform: an Efficient directional multi-resolution

image representation, IEEE Transactions on Image

Processing, vol. 14, issue 12, pp. 2091-2106.

Haralick, R.M, et al., 1973.Textural features for image

classification, IEEE Transactions on Systems, Man,

and Cybernetics, vol.3, no.6, and pp.610-621.

Karen P. Hollingsworth, Kevin W. Bowyer, 2008. The

Best Bits in an Iris Code, IEEE Transactions on

Pattern Analysis and Machine Intelligence (PAMI).

Ma, L, et al. 2003. Personal identification based on iris

texture analysis, IEEE Trans. Patt. Analy. & Mach.

Intel vol. 25, No. 12, pp. 1519-1533.

Sanchez-Avila, C. et al.,2002.Iris-based biometric

recognition using dyadic wavelet transform,IEEE

Aerospace and Electronics Systems Magazine ,

Volume: 17 Issue 10, pp. 3-6.

Wildes, R. P, 1996, et al., A machine-vision system for iris

recognition, Machine Vision and Applications,

Springer-Vela.

Zhenan Sun et al., 2005.Improving Iris Recognition

Accuracy via Cascaded Classifier, IEEE Transactions

on systems, Man and Cybernetics-part C: Applications

reviews, Vol.35, No.3.

CASIA, 2003.Chinese Academy of Sciences – Institute of

Automation, Database of 756 Grayscale Eye Images.

http://www.sinobiometrics.com Version 1.0.

Jafar M. H. Ali, Aboul Ella Hussanien,2003.An Iris

Recognition System to Enhance E-security

Environment Based on Wavelet Theory, AMO -

Advanced Modeling and Optimization, Volume 5,

Number 2.

Ma, L, et al., 2002. Iris Recognition Based on

Multichannel Gabor Filtering, Proc. Fifth Asian Conf.

Computer Vision, vol. I, pp. 279-283.

Zhu, Y. et al., 2000. Biometric Personal Identification

Based on Iris Patterns, Proc. Int’l Conf. Pattern

Recognition, vol. II,pp. 805-808.

A NOVEL TECHNIQUE FOR IRIS RECOGNITION SYSTEM

533