Two New Methods for Locating Outer Edge of Iris - Advances in Image and Graphics Technologies - page 346

the combination of the threshold segmentation and the radial gradient local maximum
values to locate the iris by using 23ms. In 2011, He Xiaofu [4] ,Tian Xuzi, Zhang
Yuan and Huang Liyu [6] improved the method of Hough transform to locate the iris
edge and the time they locate iris were 0.56s and 2.5s. The following methods also
cannot be used in real-time system either.
First, we use the canny operator and the Hough transforms to extract iris inner
edge. The threshold of Hough transform is determined by the histogram method. Af-
ter that, we observed the iris images and segmented sub-images based on the original
images which regard the pupil center as their center and the size of these sub-images
is 240*220. These images contain the entire iris image and decrease the number of
pixels. Thus the speed of locating the outer edge of iris images has been increased.
Finally, aiming at the problem of the blurring outer edge of iris, we present the circle
integral and linear segment methods to locate iris outer edge.

2 The Localization of Inner Edge of Iris

The iris inner edge is just the edge between the pupil and iris which is clear. Moreover
the inner edge can be approximately considered as a circle. And the operation of
smooth can be used to reduce noise. Therefore, we use the method of Hough circles
transform to detect and locate the pupil region.
Hough transform method is a kind of gradient method which can solve the circle
transform problem. First of all pupil area can be distinguished by applying a threshold
step and the method of diagram can help to choose an exact threshold which are used
in the image binary progress. Canny edge detector can be used at next step to get edgy
image. At last using the Hough method to estimate the pupil center and radius on the
edge image x , y . The work is described in the following equations.

Hx ,y ,r∑ hx , y ,x ,y ,r (1)

Where x and y present the coordinates of the circle centers, r presents radius of
every circle which has been detected. These three parameters constitute some parame-
ter sets and H presents an accumulator to choose candidate center by voting that pa-
rameter sets. Selecting centers which meet the accumulator condition and sorting
these centers along with the cumulative value of votes in descending order.

hx , y ,x ,y ,r 1, gx , y ,x ,y ,r0

0, gx , y ,x ,y ,r0

(2)

Where gx , y ,x ,y ,rx x y y r denotes the voting function
which satisfies the parameter equation of circle. Voting the parameter
group x ,y ,r when the edge point x , y is on the circle which consists of the pa-
rameter group x , y ,x ,y ,r . According to the method, we can get some circles of
iris image except the pupil's edge. So, we stipulate the scope of the size of the pupil
and the distance between the pupil center and the center of the image.
An example of this process on an eye image is shown from Fig. 1 to Fig.4.