18-04-2012, 04:01 PM
A NOVEL IRIS RECOGNITION METHOD BASED ON FEATURE FUSION
01380440.pdf (Size: 333.32 KB / Downloads: 51)
Introduction
Iris recognition is one of the most active research
topics for the last few decades in biometric technology
because iris pattem has stable and distinctive features for
personal identification. In addition every iris has fine and
unique pattems and does not change over time since two or
three years after the birth.
Daugman''' implemented the system with circular
edge detector for localization of iris, 2-D Gabor filter for
feature extraction, and 1024bits iris code for Hamming
distance match. Bot the method of Daogman based on local
features concentrated on ensuring that repeated image
capture produces irises on the same location within the
image, had the same resolution, and were glare-free under
fixed illumination.
Iris image preprncessing
The preprocessing stage of iris recognition is to isolate
the iris region in a digital eye image. The iris region, show
in Figure 1-a, can be approximated be two circles, one for
the iris and sclera boundary and another for the iris and
pupil boundary. The eyelashes and eyelids normally
occlude the upper and lower parts OS the iris region. In
addition, specular reflections can occur within the iris
region corrupting the iris pattem. A technique is required to
isolate and exclude these artifacts as well as locating the
circular iris region.
Iris localization
The iris localization is to detect the iris area between
pupil and sclera from an eye image. It is important to
precisely detect the inner boundary and the outer boundary
to find out that area exactly. First, we take the center of the
pupil as the reference point, and compute the distance from
the point to the boundaries as the pupil radius. Since pupil
is darker than its surroundings distinct. So we project the
image in the vertical and horizontal direction to
approximately estimate the center coordinates of the pupil.
The coordinates corresponding to the minima of the two
projections profiles are considered as the center coordinates
of the pupil. By some prior knowledge we how the center
of the iris(outer boundary) is closed to the center of pupil.
The Proposed Iris Feature Extraction Method
In this paper, we propose the adaptive method to
facilitate the ins pattem matching by fusing global features
and local features. Both the features are extracted from the
log Gabor wavelet filter at the different levels. The first one
is the global feature that is invariant to the eye image
rotation and the inexact iris localization. The statistics of
textures features is used to represent the global iris features.
The introduction of the global features has taken strengths
of decreasing the computation demand for local matching
and compensating the error in localizing iris region.
the local features extraction
The global feature represents the global characteristic
of iris image well. But the local difference can't availably
reveal and the recognition rate is affected with the different
iris having the similar global features. The global feature
need local feature to perfect the recognition. This paper
encodes the iris image into binary code to match with
hamming distance. Due to the texture at the high frequency
levels is strongly affected by noise. We extract the local iris
feature at the intermediate levels.
Conclusions
A new iris recognition system is provided in this paper.
The key point of our new method for iris recognition is the
fusion of global feature and local feature. The global
feature is extracted with the log Gabor wavelet filter that
allows arbitrarily large bandwidth filters to be conshucted
while still maintaining a zero DC component in the
even-symmetric filters. The local feature obtained by log
Gabor filter reflects the local gray density relation. Both the
fusion of features and the setup of threshold speed the
recognition system, distinctly reduce the FAR, perform
robustly with different image quality. Results achieved in
this paper are promising and some additional researches
will be performed in the future with greater amount and
variety of iris images.