Iso 9001: 2008 Certified Journal, Volume 4, Issue 7, July 2014

International Journal of Emerging Technology and Advanced Engineering 
Website: www.ijetae.com (
ISSN 2250-2459,
 
ISO 9001:2008 Certified Journal,
 Volume 4, Issue 7, July 2014)
 
960 
III. P
RESENT 
S
CHEME
A state of art iris recognition system comprises the 
following four basic modules:- 
 Image acquisition 
 Iris segmentation 
 Normalization and iris code generation 
 Comparison and recognition 
A typical iris segmentation process includes the 
following steps: demarcating the iris inner and outer 
boundaries at the pupil and sclera; demarcating its upper 
and lower eyelids if they occlude; and detecting and 
excluding any superimposed occlusions such as a eyelash, 
eyelids, shadows, or reflections. We have presented a 
method which takes less time for segmentation and finds 
accurate circles for pupil and iris. This technique uses 
median filters to enhance the image before segmentation. 
Thus segmentation is applied which divides the image into 
a number of clusters only, in which iris region belongs to 
single cluster only. After that edge detection has been 
carried out in order to find better edges needed in hough 
transform for finding iris circle. Below is the proposed 
algorithm in this work. 
Step I- Preprocess the input image in order to extract eye 
region from the input image. Filters and extract eye region 
from the input image. Filters and extract eye region from 
the input image. Filter and contrast adjustment functions 
are used in this step to better highlight eye region. 
Step II- Smoothing filters have been applied on the 
extracted eye portion in order to get better results in image 
segmentation. 
Step III- Segmentation has been done for the extracted 
portion using fuzzy means clustering and iris region has 
been located in a cluster. 
Step IV- Edge detection has been applied on the segmented 
image. We tried different filters like Sobel, Prewitt, Canny 
but Canny gave better results in edge detection. 
Step V- Circular Hough transform has been applied in order 
to get iris and pupil area from the image. 
Step VI- Inner o0f the pupil area has been excluded from 
the image as it is not a biometric tool and iris region has 
been kept as original. 
Step VII- Upper and lower eyelids have been removed 
which results in required iris neede for matching. 
Step VIII- Upper and lower eyelids have been removed 
which results in required iris needed for matching. 
Step IX- Finally template has been generated and has been 
matched to the existed database. 
V. E
XPERIMENTAL 
R
ESULTS
The software implementation of the project has been 
done using MATLAB. MATLAB stands for MATRIX 
LABORATORY, software developed by Mathsworks 
honor in USA. Procedure of the proposed algorithm has 
been explained as below: 
It comprises of few basic modules: Image Preprocessing, 
Phase -1 Iris Localization, Phase-2 Iris Localization, and 
Fitting Non-circular Contours. Literature reveals that 
circular hough transform(CHT) is tolerant to broken 
contours of the objects in the ideal images. However It may 
not be true for the non ideal data. It may be because of the 
non uniform illumination, non circular iris contours and 
occlusions such a hair, glasses, contact lens, eyelids and 
eyelashes. However to make CHT robust for the non ideal 
data as well, we augment it by the image gray level, 
statistics for example, global average gray level intensity, 
lower and upper saturated gray level limits of the eye 
image. This combination of CHT and gray level statistics 
of image results in a better strategy for finding iris. It 
implies that a circular region in an eye image would be 
considered as an iris/pupil region, provided the following 
two conditions are true: 
 A peak corresponding to the pupil/iris circle 
should be present in CHT accumulator; and 
 Gray level intensity of that circular region should 
be relatively low with respect to some threshold 
value. Next, module localizes the pupil and iris 
circles in the preprocessed eye image using an 
effective scheme. 
As we mentioned earlier, our focus is on precise 
localization of the iris inner and outer contours, therefore 
the final localized iris may contain eyelid and eyelashes 
occlusions as in. 

---