ROBUST IRIS RECOGNITION IN UNCONSTRAINED ENVIRONMENT

USING LOCAL AND GLOBAL FEATURE

GOPINATHA T
M.tech, Dept of SCN, EWIT, VTU Belgaum
gopinathktg@gmail.com

JAGADISHA N
Asst. prof Dept.of ISE, EWIT (Research Scholar) VTU Belgaum
Jagadisha.n83@gmail.com

DR. C. M. PATIL
VVCE, Mysore
cmpatil@gmail.com

ABSTRACT: The Iris recognition for the iris images taken from a distance from the sensor is a
major challenge in the biometric platform. The fragile bits that are the bits, which are not
consistent at a particular position in the iris feature bits when compared with the database iris are
taken into consideration for analysis and the Zernike transform is used for the analysis of the iris
image feature in order to extract more from the local iris region. The fragile bits are for global
feature extraction and Zernike transform is to remove local feature. This local and global feature
extraction will help in efficiently recognizing the iris with the database. The Matlab based
simulation is carried out on the proposed method and the results are tabulated and compared. The
False Acceptance Rate (FAR) and False Rejection Rate (FRR) are used for the calculation of the
accuracy of the recognition.

KEYWORDS: Biometrics, iris recognition, Zernike transform, FAR, FRR,

INTRODUCTION

The Iris recognition for the iris images taken from a distance from the sensor is a major
confrontation in the biometric podium. The fragile bits that are the bits, which are not consistent at
a particular position in the iris feature bits when compared with the database iris are taken into
consideration for analysis and the Zernike transform is used for the analysis of the iris image
feature in order to extract more from the local iris region. The fragile bits are for global feature
extraction and Zernike transform is to remove local feature. This local and global feature
extraction will help in efficiently recognizing the iris with the database. The Matlab based
simulation is carried out on the proposed method and the results are tabulated and compared. The
False Acceptance Rate (FAR) and False Rejection Rate (FRR) are used for the calculation of the
accuracy of the recognition spectrum. In order to get a well focused iris test data the algorithm
takes the concept of maximizing the spectral power by adjusting the focus of the system, or giving
the person audio feedback to adjust their position in front of the camera. More detail on the
focusing procedure is given in the appendix of [2]The main purposefulness of [3] is to reveal the
availability of the UBIRIS.v2 database, a multisession iris images database which especially
covers data taken in the visible wavelength, at-a-distance (between four and eight meters) and on
on-the-move. The procedure starts with the segmentation of the region corresponding to the iris
ring in the close-up eye image. Further, to compensate for varying pupils sizes, imaging distances,
and distortions, the data are transformed into a double dimensionless polar coordinate system,
through a process known as the Daugman Rubber Sheet [3].

There are numerous proposals to develop systems that operate in the visible wavelength and in less
constrained environments. These imaging conditions engender acquired noisy artifacts that lead to
severely degraded images, making iris segmentation a major issue. Having observed that existing
iris segmentation methods tend to fail in these challenging conditions, we present a segmentation
method that can handle degraded images acquired in less constrained conditions They offer the
following contributions: 1) to consider the sclera the most easily distinguishable part of the eye in
degraded images 2) to propose a new type of feature that measures the proportion.

LITERATURE SURVEY
The most important work in the early history of iris biometrics is that of Daugman. Daugmans
1994 patent [1] and early publications (e.g., [2]) described an operational iris recognition system in
some detail. It is fair to say that iris biometrics as a field has developed with the concepts in
Daugmans approach becoming a standard reference model. Also, due to the Flom and Safir patent
and the Daugman patent being held for some time by the same company, nearly all existing
commercial iris biometric technology is based on Daugmans work.

Daugmans patent states that The system acquires through a video camera a digitized image of an
eye of the human to be identified. A 2004 paper [3] said that image acquisition should use near
infrared illumination so that the illumination could be controlled, yet remain unintrusive to
humans. Near-infrared illumination also helps reveal the detailed structure of heavily pigmented
(dark) irises. Melanin pigment absorbs much of visible light, but reflects more of the longer
wavelengths of light.

Wildes [4] describes an iris biometrics system developed at Sarnoff Labs that uses a very different
technical approach from that of Daugman. Whereas Daugmans system acquires the image using
an LED-based point light source in conjunction with a standard video camera, the Wildes system
uses a diffuse source and polarization in conjunction with a low l\ght level camera. When
localizing the iris boundary, Daugmans approach looks for a maximum in an integro- differential
operator that responds to circular boundary. By contrast, Wildes approach involves computing an
binary edge map followed by a Hough transform to detect circles. In matching two irises,
Daugmans approach involves computation of the normalized Hamming distance between iris
codes, whereas Wildes applies a Laplacian of Gaussian filter at multiple scales to produce a
template and computes the normalized correlation as a similarity measure.

Bachoo and Tapamo [5] approach the detection of eyelash occlusion using the gray-level co-
occurrence matrix (GLCM) pattern analysis technique. The GLCM is computed for 21x21
windows of the image using the most significant 64 grey levels. A fuzzy C-means algorithm is
used to cluster windows into from 2 to 5 types (skin, eyelash, sclera, pupil, and iris) based on
features of the GLCM. However, with proper preprocessing [10, 6, 11, 12], translation and scale
invariance can be achieved. In order to make Zernike moments invariant to translation, the origin
of the unit disk is translated to the center of mass of the shape. For scale invariance, Khotanzad
and Hong [13] resampled the shape so that the area of the shape has a constant value. Bin and Jia-
xiong [7] improved the rota- tion and scale invariance property significantly by fitting the smallest
circle on the shape and used this circle as the unit disk. Zernike moments which have been

653
acquired us- ing the latter method were normalized by dividing over the shape area.

IMPLEMENTATION

The proposed work is developed considering the disadvantages of the previous works that are not
consistent with the dynamism created by the acquiring of the iris using the iris acquired from a
distance and also with many noise introduced in it.

The ZMs have been shown to constitute robust image features, which are more tolerant to noise,
information redundancy, viewpoint change, partial occlusion, etc.[8]. Such properties are highly
desirable in order to accommodate for noise and imaging quality variations as commonly observed
in the visible illumination eye images acquired under less constrained environments. These
Zernike moments will accommodate the local variations in the image details and the fragile bits
would accommodate the global variations in the image details. Such combined information can
allow us to make better decisions and benefit from the outcome of matching performance using
local texture matches which are more tolerant to variations/noise, and also the global iris texture
matches which have its strength in less noisy iris region pixels.

There are two steps in applying the proposed algorithm the first one is normalizing the iris and the
second one is the removal of both local and global features.

Figure1: Segmentation

The steps involved in the normalizing the iris procedure is:

1. Image enhancement using Retinex algorithm 2.Image reflection detection and masking, 3.
Coarse segmentation using random walker algorithm, 4.Daugman to original image,5.Boundary
refinements ---compare with Daugman and refine the output obtained from random walker.
The flow diagram of the first method is as given in Figure 1.

654
The second procedure involves the extraction of both the local and the global features by using the
feature extraction techniques like Gabor log feature and the Zernike transform. Both these feature
are considered here to have the iris recognition carried out.

The steps involved in the removal of both local and global features procedure is:

1.Feature encoding and matching strategy,2.Applying preprocessing and mapping 3.Applying log
Gabor feature extraction, 4.Zernike moments phase based encoding,5.If the combined score is
either for a particular class then decision will be true for the class.

The flow diagram of the second procedure is given as below in figure 2.

.
Figure 2.Global and Local Feature extraction and Matching

RESULTS AND DISCUSSION

The Matlab based simulation was applied on the UBIRIS database and for both the normalization
and the recognition procedures were applied. The intermediate outputs of the procedure are shown
in the following figures. The figure.3 shows the image applied with the retinex algorithm for
getting a enhanced image of the iris.

We could observe that the output of the retinex algorithm gives us the enhanced image. The
Daugman method would be applied to segment out the iris and the pupil for normalization.
Figure4 shows the output of the Daugman based segment.

The local and the global feature extraction using both the log Gabor and the Zernike transform are
carried out and the scores of each of the method is obtained to get the recognition levels of each
iris.Figure.5 Shows the Zernike polynomials diagram from the iris image while it was considered
as smaller blocks.

655
Figure 3:.Retinex algorithm output on the image Figure 4: Daugman segmentation

Figure 5: Zernike Polynomials

After getting these feature the score of both the local feature that is Zernike transform and the
global features that is the log Gabor features are considered for the iris recognition. If the score is
higher while compared to the database iris then the image would belong to that class. The False
Acceptance Ratio (FAR),VFalse Rejection Ratio(FRR) and the Total Success Rate (TSR) for the
UBIRIS database with this method was taken for analysis .Around 10 classes with 20 iris image
for each class was taken for analysis and it was found that this method gave a better TSR.

The TSR for the current technique was almost 93% as shown in the figure 6.The figure 7 depicts
the FAR vs. FRR graph in our method.

656
 Figure 6: TSR vs. Threshold Figure7. FAR vs. FRR

CONCLUSION

The Matlab based simulation on the proposed method was carried and the results were analyzed
according to the FRR , FAR and the TSR on the UBIRIS database. It was found that this method
performed better than many of the previous methods. The TSR was around 93% for 200 images
that was taken for consideration for this analysis.

REFERENCES

John Daugman. Biometric personal identification system based on iris analysis. U.S. Patent No.
5,291,560, March 1994
John Daugman. How iris recognition works. IEEE Transactions on Circuits and Systems for
Video Technology, 14(1):2130, 2004.
H. Proenca, S. Filipe, R. Santos, J. Oliveira, and L. A. Alexandre, The UBIRIS.v2: A database of
visible wavelength iris images captured on- the-move and at-a-distance, IEEE Trans. Pattern
Anal. Mach. Intell., vol. 32, no. 8, pp. 15291535, Aug. 2010.
H. Proenca, Iris recognition: On the segmentation of degraded images acquired in the visible
wavelength, IEEE Trans. Pattern Anal. Mach. Intell., vol. 32, no. 8, pp. 15021516, Aug.
2010.
J. R. Matey et al., Iris on the move: Acquisition of images for iris recognition in less constrained
environments, Proc. IEEE, vol. 94, no. 11, pp. 19361947, Nov. 2006.
A.Kumar and A. Passi, Comparison and combination of iris matchers for reliable personal
authentication, Pattern Recognit., vol. 43, no. 3, pp. 10161026, 2010.
Using Fragile Bit Coincidence to Improve Iris Recognition
A. Khotanzad and Y. H. Hong, Invariant image recognition by Zernike moments, IEEE Trans.
Pattern Anal. Mach. Intell., vol. 12, no. 5, pp. 489497, May 1990.
Chun-Wei Tan, Ajay Kumar Accurate Iris Recognition at a Distance Using Stabilized Iris
Encoding and Zernike Moments Phase FeaturesIEEE Transactions On Image Processing,
Vol. 23, No. 9, September 2014

657