EFFECTIVENESS OF FEATURE DETECTION OPERATORS ON THE PERFORMANCE OF IRIS BIOMETRIC RECOGNITION SYSTEM

International Journal of Network Security & Its Applications (IJNSA), Vol.5, No.5, September 2013
DOI : 10.5121/ijnsa.2013.5506 73
EFFECTIVENESS OF FEATURE DETECTION
OPERATORS ON THE PERFORMANCE OF IRIS
BIOMETRIC RECOGNITION SYSTEM
Binsu C. Kovoor1
, Supriya M.H.2
and K. Poulose Jacob3
1,3
Department of Computer Science, Cochin University of Science and Technology,
Cochin, Kerala, India
2
Department of Electronics, Cochin University of Science and Technology, Cochin,
Kerala, India
ABSTRACT
Iris Recognition is a highly efficient biometric identification system with great possibilities for future in the
security systems area.Its robustness and unobtrusiveness, as opposed tomost of the currently deployed
systems, make it a good candidate to replace most of thesecurity systems around. By making use of the
distinctiveness of iris patterns, iris recognition systems obtain a unique mapping for each person.
Identification of this person is possible by applying appropriate matching algorithm.In this paper,
Daugman’s Rubber Sheet model is employed for irisnormalization and unwrapping, descriptive statistical
analysis of different feature detection operators is performed, features extracted is encoded using Haar
wavelets and for classification hammingdistance as a matching algorithm is used. The system was tested on
the UBIRIS database. The edge detection algorithm, Canny, is found to be the best one to extract most of
the iris texture. The success rate of feature detection using canny is 81%, False Accept Rate is 9% and
False Reject Rate is 10%.
KEYWORDS
Iris, Canny, Daugman, Prewitt, Zero Cross, Sobel
1. INTRODUCTION
The contemporary advances of information technology and the growing requirements for security
in an interlinked society, has created a huge demand for intelligent personal identification system.
In addition to this, as people become more connected electronically, the capability to accomplish
an exceedingly precise automatic personal identification system is considerably more critical. For
identification and veriﬁcation ofindividuals, to controlaccess to the secured areas or materials,
technologies that exploit biometrics have great potential. Biometric identification refers to the
identification of humans by their distinctive measurable physiological or behavioral
characteristics [1]. Since many of these are unique to an individual, biometric identifiers are
fundamentally more dependable and it relates a person with an earlier recorded identities based on
how one is or what one does.
An ideal biometric should possess four characteristic, namely, universality, uniqueness,
permanent and collectability. Universality means each person should possess the characteristic;
uniqueness means no two persons should share the characteristic;permanentis that the
characteristic neither should change nor be alterable and collectablemeansthe characteristic is
readily presentable to a sensorand is easily quantifiable.Facial imaging, hand and finger

74
geometry, eye-based methods,signature, voice, vein geometry, keystroke, and finger- and palm-
print imaging etc. are some of the currently pursued biometric traits in security system.
Iris can be used as an optical biometric trait for identifying a personsince it has a highly detailed
pattern that is unique and permanent [2][3][4]. Arching ligaments, furrows, ridges, crypts, rings,
corona, freckles and zigzag collarette[2] are the various unique features of the iris. These features
provide extraordinary textural patterns that are distinctive to each eye of an individual and are
even distinct between the two eyes of the same individual[4].
The image of eye comprises of iris, pupil, sclera, eyelid and eyelashes. The detection of the iris
from the eye image can be performed by segmenting the annular portion between the pupil and
sclera.Iris recognition techniques identify a person by mathematically analyzing the unique
patterns of iris and making comparisons with an already existing knowledge base. The overall
performance of iris recognition system is decided by the accuracy of conversion of iris features
into iris code. The UBIRIS database is used to implement and test the model for the iris
recognition system. [5]
2. METHODOLOGY
The unique iris pattern from a digitised image of the eye is extracted and encoded into a biometric
template using the image processing techniques. This can later be stored in the knowledge base.
The unique information in the iris is represented as objective mathematical representation. This is
checked against templates for resemblances. When a person wishes to be authorised by an iris
recognition system, their eye has to be first photographed, and a template is created for their iris
region. The template is compared with the other templates in the knowledgebase. The comparison
can be made till a matching template is found and the person is recognized, or no match is found
and the person is overruled.
There are five main steps for the iris recognition process. The first step is the enrolment, where
the eye image is captured. The next step is the segmentation of the iris from the other parts of the
eye image. Normalization is the third step, in which the iris pattern is scaled to a constant size.
Iris is represented as iris code in the fourth step. The classification phase is the final step, where a
matching technique is used to find out the similarity between the two iris code.Figure1 depicts the
schematic for an Iris recognition system.
Figure 1. Schematic of Iris Recognition system
2.1. Iris Segmentation
The upper and lower parts of the iris region are obstructed by the eyelids and eyelashes. Iris
pattern can also get corrupted by the specular reflections within the iris region.The artefacts are
separated and the circular iris region is detected.The quality of eye imageswill greatly influence
the success of segmentation.Thus a good segmentation algorithm involves two procedures, iris
localization as well as noise reduction. For finding the boundary between the pupil and iris, as
well as the boundary between the iris and the sclera of the acquired image, iris localization
process is used.In the localization process, the noises (non-iris parts) are removed from the
acquired image and this process is referred as noise reduction. Pupil, sclera, eyelids,
Imageacquisi
tion
Iris
Segmentation
Normalisation
& Unwrapping
Feature
Extraction
Classification

75
eyelashes,and artefacts are the noises in the acquired image [6]. The flow diagram of the iris
segmentation is as depicted in Figure 2.
Figure 2. Steps of Iris segmentation
The centre pixel is determined for the image obtained from the noise reduction process. A circular
strip of the iris image is obtained based on the centre co-ordinates of the pupil. For detecting the
inner and outer boundary ሺ‫ܤ‬௜௡ ௢௥ ௢௨௧ሻ of the iris, Integro-Differential operator [4] was used. The
integro-differential operator is shown in equation (1).
‫ܤ‬௜௡ ௢௥ ௢௨௧ ൌ ݉ܽ‫ݔ‬ሺ௥,௫೚,௬బሻ ቤ‫ܩ‬ఙሺ‫ݎ‬ሻ ൈ
߲
߲‫ݎ‬
ර
‫ܫ‬ሺ‫,ݔ‬ ‫ݕ‬ሻ
2ߨ‫ݎ‬௥,௫೚,௬బ
݀‫ݏ‬ቤ (1)
In the above equation, the eye image is denoted by ‫ܫ‬ሺ‫,ݔ‬ ‫ݕ‬ሻ, the radius to search for by ‘r’,
Gaussian smoothing function by ‫ܩ‬ఙሺ‫ݎ‬ሻ and s is the contour of the circle given by r,x0,y0. By
changing the radius and centrex and y position of the circular contour, the operator explores for the
circular path to identify the maximum change in the pixel value. In order to accomplish precise
localization, the operator is applied iteratively.
2.2. Iris Normalization and Unwrapping
After segmenting the iris region from an eye image, it is transformed to fixed dimensions so that
comparisons can be easily made. Transforming iris into polar coordinates is known as
unwrapping process. Iris region will be mapped into a constant dimension by the normalization
process, and thus two photographs of the same iris under different conditions will have same
characteristics features. Figure 3 illustrates a normalised iris image.

76
Figure 3. Normalized iris image
2.3. Feature Extraction
In the process of feature extraction, the most distinguishing information present in an iris pattern
is extracted. The substantial features of the iris are encoded so that evaluation between templates
can be done. This helpsin the precise identification of the individual.
Initially, histogram equalization is carried out to enrich the iris texture in the normalized
image.The canny edge detector [7] is used to extract the iris texture from the normalized image.
For reducing the dimension, the 2D image of the edge is detected and transformed into a 1D
energy signal by the use of vertical projection. A set of low frequency and high frequency
coefficients are obtained by applying discrete wavelet transform to these 1D energy signals. The
low frequency coefficients, which have a dimension of 64 bytes, can be considered as the iris
templates. The high frequency coefficients do not comprise of any significant information and
hence can be omitted. Figure 4 depicts the different steps associated with the feature extraction
stage.
2.3.1. Histogram Equalization
Through histogram equalisation the local contrast of many images can be significantly improved
and the intensities can be better distributed on the histogram. This allows for areas of lower local
contrast to gain a higher contrast without affecting the global contrast and is accomplished by
effectively spreading out the most frequent intensity values. Figure 5 shows the image attained
after histogram equalization. The occlusion of the eyelid is represented as domes in the
unwrapped image.
Figure 5. Histogram Equalised Image
Histogram
equalisation
Edge
detection
Vertical
Projection
DWTNormalized
image
Iris
Features
Figure 4. Feature Extraction Stages

77
2.3.2. Edge Detection
Edge Detection is performed using the classical operators which are classified into three main
categories vis a visGradient, Laplacian Based and Canny. The Gradient- based operator include
Roberts[8], Sobel[9] and Prewitt[9] edge detection operators while the Laplacian-based include
LOG[9] and Zero Cross edge[9] detection operator and the Canny edge detector [9][10]. Iris
texture in the normalized image will be improved by histogram equalization process. By the use
of these edge detection operators, the iris texture is extracted and comparative study is carried out.
It is found that canny edge detection technique is able to obtain most of the iris texture from the
enhanced image.
2.3.3. Vertical Projection
Vertical projection reduces the system complexity by converting the 2D signal to 1D signal. For
vertical projection, energy of each row of the edge detected image is calculated and is converted
into a row vector. The generalized equation is shown in equation (2).The dimension of
normalized image is ݉ ൈ ݊and is taken as128 ൈ 512. Hence, after vertical projection its
dimension is m, which is equal to 128.
൥
‫ݔ‬ଵଵ … ‫ݔ‬ଵ௡
‫ڭ‬ ‫ڰ‬ ‫ڭ‬
‫ݔ‬௠ଵ … ‫ݔ‬௠௡
൩ ՜ ൥෍|‫ݔ‬ଵ௜|ଶ
௡
௜ୀଵ
… ෍|‫ݔ‬௠௜|ଶ
௡
௜ୀଵ
൩ (2)
2.3.4. Discrete Wavelet Transform
The discrete wavelet transform (DWT) divides the signal into mutually orthogonal set of wavelets
[11]. The signal x is passed through a series of filters and DWT is calculated. Initially a low pass
filter with impulse response g[n] is used to pass the samples, which results in to a convolution and
is given in equation (3).
ܻሾ݊ሿ ൌ ሺ‫ݔ‬ ‫כ‬ ݃ሻሾ݊ሿ ൌ ෍ ‫ݔ‬ሾ݇ሿ ݃ሾ݊ െ ݇ሿ
ஶ
௞ୀିஶ
(3)
A high pass filter h[n] is also used simultaneously to pass the signal. The outputs from the high
pass filter give the detail coefficientsand low pass filters give the approximation coefficients.
Thedimensions of the coefficients are 64 bytes each, since the dimension of 1D signal is 128
bytes. These two filters are referred as quadrature mirror filter. Here Haar wavelet [12] is used for
wavelet transform. After wavelet transform, a set of low frequency coefficients and high
frequency coefficients each of dimension 64 bytes is obtained. After DWT, it is observed that
approximation coefficients contain information and detailed coefficients do not have any
information. Hence approximation coefficients with dimensions 64 bytes is selected as feature
vector and stored in database.
2.4. Classification
In recognition stage the features of the input eye image is compared with that of the features that
is already stored in the database and if it matches, the corresponding eye image is identified
otherwise it remains unidentified. Since a bitwise comparison is necessary Hamming distance was
chosen for identification

78
2.4.1. Hamming Distance
The Hamming distance [13] gives a measure ofdifference between two bit patterns‫ݔ‬௞and‫ݕ‬௞. In
the classification stage, the comparison of iris code is done by hamming distance approach.
Hamming distance D is given by equation (4).
‫ܦ‬ ൌ
1
݊
෍ ‫ݔ‬௞
௡
௞ୀଵ
⊕ ‫ݕ‬௞ (4)
where, x and y are the two bit patterns of the iris code. n indicates number of bits. Hamming
distance D gives out the number of disagreeing bits between x and y.
Ideally, the hamming distance between two iris codes generated for the same iris pattern should
be zero; however this will not happen in practice due to fact that normalization is not perfect. The
larger the hamming distances (closer to 1), the more thetwo patterns are different and closer this
distance to zero, the more probable the two patterns are identical. By properly choosing the
threshold upon which we make the matching decision, one can get good iris recognition results
with very low error probability.
3. RESULTS AND DISCUSSIONS
The system was tested using the UBIRIS database [5] which included 1877 images from 241
persons collected in two sessions. The images collected in the first photography session were low
noise images. On the other hand, images collected in the second session were captured under
natural luminosity factor, thus allowing reflections, different contrast levels, and luminosity and
focus problems thus making it a good model for realistic situations.
Fifty sets of eye images from UBIRIS database was taken for identification. Each set consists of
three eye images of a person taken at different time. From each set a single eye image was
randomly selected and its features were stored in the database. Therefore a total of 50 images
were used for simulation. These images are called registered images since its feature is stored in
the knowledge base. The main challenge in the identification is to identify the other two images in
each set whose features are not stored. 50 images whose features are not stored in the database are
also used to test the algorithm. These images are called as unregistered images. An efficient
algorithm should identify all registered images and reject all unregistered images. Performance of
iris acceptance algorithm is validated using four parameters -False Reject (FR), False
Accept(FA), Correct Reject(CR) and Correct Accept(CA). FR is obviously the case where we
judge a pattern as not the target one while it is. FA is when the pattern is considered as the target
one while it is not. CR is when the pattern is correctly judged as being not the target one. Finally,
CA is when the pattern is correctly considered to be the targeted one. These outcomes are
illustrated in Figure 6. It was found that an optimum result is obtained at hamming distance
threshold of 0.4. If hamming distance, between the iris code in the knowledgebase and the testing
iris code, is less than the threshold then the person is authentic and accepted otherwise rejected as
imposters.
Using MATLAB, a comparison study between different classical operators, Canny, Sobel,
Prewitt, Roberts, log and zero cross was also done. The operators were applied to the enhanced
normalized image. The results, presented in Figure 7, show the performance of each of the
operators. It was found that the Canny operator outperforms the others; in fact it was the only
operator which was able to extract most of the iris texture.The success, false acceptance and false

79
rejection of the Iris recognition are recorded for various edge detection operators -Prewitt, Robert,
Sobel, Zero Cross, Log and Canny system. The statistical details of the success ratio (SR), false
acceptance ratio (FAR) and false rejection ratio (FRR) of Prewitt, Robert, Sobel, Zero Cross, Log
and Canny edge detection operators are given in Table1. The mean value success ratio of Canny
is greater than the other operators. Similarly the success ratio of Zero Cross and Log operators are
greater than Prewitt, Robert and Sobel. The mean value of false acceptance ratio and false
rejection ratio of Canny is less than other operators. The performance of an operator is said to be
acceptable when the magnitude of SR is high and magnitude of FAR or FRR is low. In this
study, it is found thatCanny edge detection technique is the most efficient to capture the iris
texturewhen compared to the other operators.
Figure 6.Decision making in iris biometric system
0
1
2
3
4
5
6
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
ProbabilityDensity
Hamming Distance
2
4
3
1. Correct accept rate
2. False accept rate
3. Flase rejectrate
4. Correct rejectratethreshold
Authentics Imposters
1

80
Figure 7.Detection of iris edge using various operators
Table 1. The statistical details of SR, FAR and FRR of Iris Identification System using various edge
detection operators
Iris System
Performance
Parameters
Statistical
Parameter
Edge Detection Operators
Prewitt Robert Sobel
Zero
Cross Log Canny
Success ratio
Mean 0.54 0.56 0.54 0.67 0.67 0.81
SD 0.06 0.09 0.07 0.06 0.06 0.07
False
acceptance ratio
Mean 0.23 0.23 0.25 0.18 0.16 0.09
SD 0.24 0.25 0.26 0.18 0.18 0.10
False rejection
ratio
Mean 0.24 0.21 0.21 0.15 0.17 0.10
SD 0.25 0.22 0.23 0.16 0.18 0.12
4. CONCLUSIONS
In the proposed system the iris region is segmented from an eye image using Daugman’s integro-
differential operator. Then it is normalised in order to counteract the imaging inconsistencies
using Daugman’s polar representation. From the obtained normalised iris pattern the features are
extracted using edge detection operators such as Prewit, Sobel, Robert, Log, Zero crossing and
Canny. Statistical analysis is performed in order to test how well each operators generate the iris
code which can be identified against a database of pre-registered iris patterns. The algorithm is

81
tested against 100 eye images out of which 50 are registered and other 50 are unregistered. The
results of the present study indicate that the canny operator is best suited operator to extract
features of iris for comparison.The success rate of feature detection using Canny is found to be
81%, False Acceptance Rate is 9% and False Rejection Rate is 10%.Hence it may be concluded
that model proposed in this study is effective for segmentation and classification of iris with less
loss of features.This algorithm can be further developed in the future for iris image capture from
the moving face.
REFERENCES
[1] Jain A, Bolle R and Pankanti S, (1999) “Biometrics: Personal Identification in a Networked
Society”, Eds. Kluwer Academic Publishers, pp1-41.
[2] Johnson R G, (1991) “Can Iris Patterns be used to Identify People?” Chemical and Laser Sciences
Division, LA-12331-PR, LANL, Calif. .
[3] Kong W, Zhang D, (2001) “Accurate Iris Segmentation based on Novel Reflection and Eyelash
Detection Model”, Proceedings of 2001 International Symposium on Intelligent Multimedia,
Video and Speech Processing, Hong Kong, pp. 263–266.
[4] Daugman J.(2004)“How iris recognition works”, IEEE Transactions on Circuits and Systems for
Video Technology, Vol.14, No.1, pp.21-30.
[5] Proenc H, and Alexandre L,(2004) “UBIRIS: Iris Image Database”, Available: http://iris.di.ubi.pt.
[6] Zhang D.(2003) “Detecting eyelash and reflection for accurate iris segmentation”, International
journal of Pattern Recognition and Artificial Intelligence, Vol. 1, No.6, pp.1025-1034.
[7] Kovesi P, “MATLAB Function for Computer Vision and Image Analysis”, Available
at:http://www.cs. uwa. edu.au/ ~pk/Research/ MatLabFns/ index.html.
[8] http://homepages.inf.ed.ac.uk/rbf/HIPR2/roberts.htm
[9] http://euclid.ii.metu.edu.tr/~ion528/demo/lectures/6/2/2/index.html
[10] Canny J,(1986) “A computational approach to edge detection”, IEEE Transaction on Pattern
Analysis and Machine Intelligence, Vol. 8, pp. 679-698.
[11] Boles 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, pp. 1185-1188.
[12] Lim S, Lee K., Byeon O, and Kim T,(2001) “Efficient iris recognition through improvement of
feature vector and classifier”, ETRI Journal, Vol. 23, No. 2, pp. 61-70.
[13] Daugman J,(1993) “High confidence Visual Recognition of Person by a test of statistical
Independence”, IEEE transactions on pattern analysis and machine intelligence, Vol.15 No.11, pp.
1148-1161.

82
Authors
Binsu C. Kovoor is working as Assistant Professor in Information Technology, Cochin
University of Science and Technology 2000 onwards. Her areas of interest include biometric
security system, pattern recognition, database systems, data mining, data structures, streaming
audio and video signals etc. She has a Life membership in ISTE and IE.
Dr. Supriya M. H joined as a faculty in the Department of Electronics, Cochin University of
Science & Technology in 1999. Her fields of interests are Target Identification, Signal
Processing, Bioinfomatics, Steganography and Computer Technology. She has presented
papers in several International C onferences in Europe, USA, and France. She is actively
involved in Research and Development activities in Ocean Electronics and related areas and
has a patent and about 87 research publications to her credit. She is a Life member of IETE and ISTE.
Dr. K.Poulose Jacob, Professor of Computer Science at Cochin University of Science and
Technology since 1994, is currently Director of the School of Computer Science Studies. His
research interests are in Information Systems Engineering, Intelligent Architectures and
Networks. He has more than 90 research publications to his credit. He has presented papers in
several International Conferences in Europe, USA, UK, Australia and other countries. Dr.
K.Poulose Jacob is a Professional member of the ACM (Association for Computing Machinery) and
a Life Member of the Computer Society of India.

EFFECTIVENESS OF FEATURE DETECTION OPERATORS ON THE PERFORMANCE OF IRIS BIOMETRIC RECOGNITION SYSTEM

More Related Content

What's hot (19)

Viewers also liked (20)

Similar to EFFECTIVENESS OF FEATURE DETECTION OPERATORS ON THE PERFORMANCE OF IRIS BIOMETRIC RECOGNITION SYSTEM (18)

Recently uploaded (20)

EFFECTIVENESS OF FEATURE DETECTION OPERATORS ON THE PERFORMANCE OF IRIS BIOMETRIC RECOGNITION SYSTEM