Graphics Reference
In-Depth Information
This method can be prominent the iris image without the whole eyebrows and
some parts of eyelids. The sub-image had narrowed down to the 1/6 size of the origi-
nal image.
4
The Localization of Outer Edge of Iris
The difference between iris and pupil pixels is not clear when the iris images in gray-
scale map. Marginalized by canny operator, only the parts of pupil is obvious. No
matter how many times one needs to change the threshold, the edge of iris cannot be
observed without any other operation.
Fig. 6.
Edge image (a)
Fig. 7.
Edge image(b)
Two of the above pictures are edge images after the use of canny edge detector.
Figure 6 uses a smaller threshold value than figure 7, but both of them cannot show
any part of the outer edge. Therefore, we cannot detect the outer edge by using the
simple method and this texture offer two methods which are circle integral and linear
segment to solve the problem.
4.1
Circle Integral
According to the feature that the iris and the pupil can be regarded as two concentric
circles, the center and radius of pupil can be used to identify the outer edge of the iris.
Searching in the b*b neighborhood of pupil center, one realizes that the scope of
search radius is between min(r) and max(R). This process is described below:
SumPixel
,R
∑∑
P
1,6
(3)
min
r
x
x
y
y
R
(4)
Where
x
,y
denotes the coordinate of the neighborhood of pupil center, and
x
,y
denotes the coordinate of the image. Where m and n denote the satisfied
following formula.
Calculating the absolute distance of
SumPixel
:
|SumPixel
SumPixel
|
(5)
Search WWH ::
Custom Search