219
A SIMPLIFIED FUSION METHOD BASED ON SYNTHETIC VARIABLE RATIO
Pang Xinhua*, Xi Bin , Chen Luyao, Pan Yaozhong,, Zhuang Wei
State Key Laboratory of Earth Surface Processes and Resource Ecology, Beijing Normal University; College of
Resources Science & Technology, Beijing Normal University, Beijing 100875, P. R. of China
-(xhpang, binxi, lychen ,pyz, terfil)@ires.cn
KEY WORDS: Remote Sensing, Data Fusion, SVR, SSVR
ABSTRACT:
A Simplified Synthetic Variable Ratio (SSVR) fusion method is presented to merge high spatial resolution panchromatic (Pan)
image and high spectral resolution multispectral (MS) images based on a simulation of the panchromatic image from the
multispectral bands. Landsat7 ETM+ images were used to assess the effectiveness of classification-oriented SSVR method in
comparison to Principal Component, multiplicative, Brovey transform and ISVR methods. Compared to other fusion methods, the
images generated by SSVR method have more information and high spatial resolution while maintaining the basic spectral
characteristic of the original multispectral image, and SSVR method is simpler to carry out than other SVR methods.
1. INTRODUCTION
With the rapid group of the internet and other electronic sources
of information, the problem of the coherent merging of
information from multiple sources has become an important
issue. This problem has many manifestation ranging from data
mining to information retrieval to multi-sensor fusion (Ronald,
2004). For many applications the information provided by
individual sensors are incomplete, inconsistent, or imprecise
(Varshney, 1997; Hall et al, 1997; Pohl et al., 1998).
Additional sources may provide complementary data, and
fusion of different information can produce a better
understanding of the observed site, by decreasing the
uncertainty related to the single sources (Farina et al., 1996;
Cl ement et al., 1993).
In data fusion the information of a specific scene acquired by
two or more sensors at the same time or separate times is
combined to generate an interpretation of the scene not
obtainable from a single sensor. Alternatively, data fusion is
done to reduce the uncertainty associated with the data from
individual sensors. Relaxing this operational definition slightly,
also the combination of the information acquired by the same
sensor at different times to improve interpretation is considered
as data fusion (Tax et al., 1997). Image fusion is used to merge
images of different spatial and spectral resolutions to create a
high spatial resolution multi-spectral combination. High
spectral resolution allows identification of materials in the
scene, while high spatial resolution locates those materials.
The actual fusion process can take place at different levels
(pixel-level, feature-level and decision-level) of information
representation (Pohl et al., 1998). Which level to choose is
determined by the purpose. For example, pixel-level fusion is
appropriate for land use classifications; because pixel-level
fusion can maintain more spectral characteristics of the original
multi-spectral image and decreasing the obscurity of image
interpretation. The common pixel-level fusion methods are
PC A, Multiplicative, Brovey Transform, HPF Transform, HIS
Transform, HDF Transform and wavelet Transform. But these
methods may cause spectral distorting and are difficult to
accomplish.
Synthetic Variable Ratio(SVR) fusion method was presented by
Munechika et al., improved by Zhang(Zhang, 1999; Zhang,
2001). But it is still difficult to calculate. The main purpose of
the paper is to present a simplified fusion method with physical
meaning based on SVR. The main purpose of the paper is to
present a simplified fusion method with physical meaning based
on SVR.
2. SSVR FUSION METHOD
2.1 SVR (Synthetic Variable Ratio) Fusion Method &
Improvement
SVR is proposed by Munechika et al (1993) taken example of
TM (30m) - SPOT (10m), and the formulation is:
XSP i = Pan H
(1)
here XSP' means band i grey value of high spatial resolution
image after fusion, Pan H means grey value of original high
resolution spatial panchromatic image, XS Li means band i
gray value of original low spatial resolution multispectral image,
and Pan, means grey value of high resolution spatial
Lsyn
panchromatic image that are synthesized by band 1, 2, 3, 4 of
multispectral TM image.
* Pang Xinhua, male, Master, major in GIS&RS, College of Resources Science & Technology, Beijing Normal University,
Beijing ,China. Recent research interests in the applications of remote sensing in crop planting area measurement and land use/cover
change detection. Tel: 010-58805750, E-mail: xhpang@ires.cn
