You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Mapping without the sun
Zhang, Jixian

Shao Yongshe a Chen Ying d Li Jing b
a Survey Department of Tongji University, No. 1239 Si Ping Road, Shanghai, China, shaoysh@sina.com
b Xi’an Research Institute of Survey and Mapping, Middle No.l YanTa Road, Xi’an, China, LiJing@sina.com
KEY WORDS: SAR image, Optical image, Lifting scheme wavelet, Image fusion, Feature extraction
Image fusion is an important kind of means improving on the targets detection and recognition in the monochromatic images. Firstly,
the paper has researched how a right wavelet base can be chosen, and decomposes the SAR image and optical image using the lifting
scheme wavelet. Then, the low-frequency images are integrated with the weighted average ways after the SAR low-frequency image
is filtered using the regional average means. In order to integrate the high-frequency images well, a kind of the fusion methods that
the windows can be changed according to the features for the noise reduction is present to improve the targets detection and
discrimination. Finally, the fusion image is used to extract the features. The result shows that the fusion image is better than the
original SAR image for feature extraction.
In recent years, the technique of radar imaging made a rapidly
development in China. Radar imaging has many outstanding
merits including the capability to work in all-time and all
weather, high spatial resolution and strong penetrability
through the ground or vegetation etc. Therefore, radar imagery
has been widely used in many fields. Differing from traditional
ways to space remote sensing, radar imagery shows the special
characteristics for the ground targets imaged. Speckle noise in
radar imagery makes it difficult to extract the edge profile of
targets using traditional ways. Integrating the SAR and optical
images has been becoming a kind of primary ways for targets
extraction and recognition.
The image fusion technique, which can integrate the different
sensors characteristics, is a kind of the ways to utilize the
images of the different sensors effectively. Comparing to the
traditional remote sensing modes, radar imaging possesses the
capability to work in all-time and all weather, high spatial
resolution and strong penetrability through the ground or
vegetation, whereas the optical imaging expresses the spectrum
information well. Accordingly, it is advantaged to integrate the
SAR image and optical image for the targets detection and
recognition Because of the different image theory and
spectrum characteristic for SAR image and optical image, the
traditional fusion methods, example for HIS, PCA and HPF etc.,
cannot make the satisfying results. The recent years, it has been
shown that the wavelet transform is a good ways for the image
fusion [7] . On the basis of having chosen an appropriate wavelet
base, in order to make it more potent to integrate the SAR
image and optical image, the paper focuses on the wavelet
high-frequency sub-images and presents a kind of the fusion
methods that the windows can be changed according to the
features for the noise reduction. The result shows that the
method may make the features keep better continuity in the
fusion image than the SAR image, which improves the effect to
detect and recognize the targets.
The wavelet base makes an important influence on the image
fusion [2][3] . The different wavelet bases demonstrate the unlike
characteristics that include the orthogonality, symmetry, filter
length, the tighten support length and the vanishing moments.
The wavelet coefficients, which relate to the wavelet base, are
the foundation of the image fusion based on the wavelet
transform. Therefore, the different wavelet bases may make the
diverse results for the image fusion.
Orthogonality presents the integrality of the wavelet base, and
simplifies the calculation process. Haar wavelet shows the trait
of the linear phase and can reconstruct the image completely,
whereas others don’t. The tighten support characteristic
presents the localization of the wavelet base. The wavelet that
doesn’t reveal the tighten support trait will lead to the error of
the truncation when the wavelet is used to decompose and
reconstruct the image. The vanishing moment determines the
astringency of the wavelet function. For the image embodying
the abundant texture, the great vanishing moment will
transform the great wavelet coefficients more. On the contrary,
it will transform the small wavelet coefficients more. The
greater the vanishing moment is, the longer the filter, which
increases the operand. Symmetry shows the linear phase, and
reconstructs the image’s edge well. Symmetrical wavelet base
can obtain the better visual effect because human isn’t sensitive
to the error of the symmetry.