206 CHINESE OPTICS LETTERS / Vol. 7, No. 3 / March 10, 2009
New enhancement of infrared image based on
human visual system
Tianhe Yu (
uuu
UUU
ààà
)
1,2∗
, Qiuming Li (
ooo
¢¢¢
²²²
)
3
, and Jingmin Dai (
µµµ
¬¬¬
)
1
1
Department of Automation Measurement and Control Engineering, Harbin Institute of Technology, Harbin 150001
2
Computer Center, Harbin University of Science and Technology, Harbin 150080
3
School of Astronautics, Harbin Institute of Technology, Harbin 150001
∗
E-mail: ythaa@163.com
Received September 18, 2008
Infrared images are firstly analyzed using the multifractal theory so that the singularity of each pixel
can be extracted from the images. The multifractal spectrum is then estimated, which can reflect overall
characteristic of an infrared image. Thus the edge and texture of an infrared image can be accurately
extracted based on the singularity of each pixel and the multifractal spectrum. Finally the edge pixels are
classified and enhanced in accordance with the sensitivity of human visual system to the edge profile of
an infrared image. The experimental results obtained by this approach are compared with those obtained
by other methods. It is found that the proposed approach can be used to highlight the edge area of an
infrared image to make an infrared image more suitable for observation by human eyes.
OCIS code: 100.0100.
doi: 10.3788/COL20090703.0206.
Histogram equalization is an algorithm most commonly
used to enhance infrared images. This algorithm can
effectively enhance the contrast of an infrared image.
However, it also causes an excessive enhancement of de-
tails and the occasional loss of edge details and features,
which undoubtedly have an adverse effect on the visua l
effects of infrared image s. The information contents are
different in different areas of an image, and so is their
impo rtance. In reality, it is not the whole image that con-
tains the same important information for an observer,
and it is more often tha t the observer is interested in
part of the imag e only. It is usually human eyes that
evaluate the final results of image enhance ment, and so
a very good visual effect can be obtained if e nhancement
of infrared images is based on a human visua l system
(HVS).
Most of the image enhancements used at present are in
the space or fre quency domain a nd few are fractal based
ways. According to the fractal theory, fractal is a more
common and typical phenomenon over a natural surface.
Like many natural images, infrared images have much
randomness in structure and noise. The gray scale of
an imag e is a coa rse surface suitable fo r the description
and analysis of fractal when it is taken as the height of
a plane. Xia et al. have success fully applied fractal to
geographic simulation, gra in analysis, seg mentation, a nd
identification of images
[1,2]
. In this letter, we present the
application of fractal in the field of image enhancement.
The dimensions of fractal re present the similarity of im-
ages, not the gray scale, but the relationship b e tween
one point and the surrounding points. When the fractal
dimension is used to enhance an image, the different
gray segments in the time domain and several frequency
ranges in the frequency domain are considered.
A variety of visual masking e ffects have been dis c ov-
ered through physiologic and psychologic researches on
the vision of human eyes
[3,4]
. As shown in Fig. 1, the
analysis of contrast sensitivity function indicates that hu-
man eyes are much mo re sensitive to the changes in the
low frequency area than those in the hig h frequency area.
From the viewpoint of space frequency, human eyes are
a low-pass linear system with limited scene dis tinguis h-
ing capacity and HVS is not sensitive to an excessively
high frequency. The perceptual experiments on the per-
ception of human eyes indicated that the attention of
human eyes is always concentrated on the fragmenta-
tion and characteristic profile of the object with all the
ordinary details totally neglected when human eyes are
sensing a natur al object. In Fig. 1, H(w), which is a nor-
malized parameter, is the sensitivity function of v ision
frequency response.
It is known through psychologic researches that an im-
age consists of three compo nents with different percep-
tual significances: smooth area, edge area, and texture
area. An area is called a smooth area if the change in its
gray scale is gentle, and an area is called an edge area
if the change in its gray scale is severe. The density of
edges in an edge area is the basis for differentiation of
an edge are a from a veined area. Images with different
information contents can thus be s imply classified ac-
cording to HVS.
It can therefo re be concluded that human eyes ar e
Fig. 1. Function of vision frequency response.
1671-7694/2009/030206-04
c
2009 Chinese Optics Letters