The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
/=i 1=1 i=i
¡=1 i=i i=i
+ a„
1=1 1=1 1=1
[R',G',B'] = F-'(IC[,IC 2 ,IC[)
IC2
1C 3
(12)
(13)
4. EXPERIMENTAL RESULTS AND COMPARISONS
4.1 Synthetic Datasets and Real MS-PAN Datasets
The main aim of this research is to determine the efficiency of
new algorithm based on ICA for merging images with a
particular resolution ratio. Due to the difficulties in obtaining
adequate imagery with particular ration, Yocky’s approach
(Yocky, 1996)is employed to synthesize some MS-PAN
datasets with particular ratio. In this approach, a Landsat TM
test image was available in the three bands,i.e., B1 (green), B2
(red), and B3 (near infrared). The image was used to synthesize
a perfectly overlapped panchromatic band at 20 m, which is
shown in fig2.
4.2 The Quality Analysis of the Fusion Image
We have adopted some quantification metrics to evaluate the
fusion quality, including entropy, mean, and standard
deviation, Average gradient. Among these metrics, entropy
explores the information changes, and an image has more
information when the entropy is bigger. And some other metrics,
such as mean, employed to evaluate the aberrance of the
spectral information. The mean calculates the degree of the
spectral information change. In our research, different
decomposition levels for wavelet have been tested. Limited by
space, only the result for true data set is show in Fig 4.
For visual analysis, we could find that our method can enhance
the image spatial resolution to a certain degree, which will
benefit those applications which are demanding strictly on the
a. original MS
bands( 128x128)
b. 2 level
c.3 level d.4 level
Fig 4 he fusion result with different decomposition levels for
true data size(512><512)
details of an image, such as image interpretation, special
cartography, and photogrammetric survey, etc. With
decomposition level increasing, more panchromatic band
information is injected into three multispectral bands and gray
levels of images seems no demonstrate change, which means
that our method may be not sensitive to wavelet decomposition
level.The synthesized data has shown the same trend. So when
we consider the computation efficiency, the less decomposition
level such as 2 or 3 is preferable. We have also adopted some
quantification metrics to evaluate the fusion quality. The
statistical data for true data is shown in table 1. It can be
included from tables: the information in both of the two datasets
is increased in the case of the injection of the
information .Because the high frequency information in the
multispectral bands is substituted by the corresponding parts in
the panchromatic band. However when decomposition level
increased, the result had no demonstrable change. It means that
our method is not sensitive to decomposition level as tradition
MRA based method. So we proposed a useful fusing algorithm.
Metrics
Band
Original
spectral
2
3
4
R
bands
113.59
113.23
113.23
113.20
Mean
G
100.23
99.87
99.87
99.84
B
93.85
93.42
93.42
93.45
Standard
deviation
R
73.50
74.10
74.09
73.78
G
70.51
73.27
73.28
72.49
B
70.65
69.44
69.44
69.71
Average
gradient
R
14.57
21.97
21.97
21.71
G
14.66
20.15
20.15
17.57
B
14.39
16.21
16.21
17.38
R
4.97
7.90
7.90
7.9
Entropy
G
4.94
7.83
7.83
7.85
B
4.85
7.80
7.80
7.80
Table. 1 statistical data for true MS-PAN dataset
5. CONCLUSION AND PROSPECTS
A new multispectral and panchromatic band merging method is
provided by combining ICA transform with discrete wavelet.
The experiment result shows that the method can improve the
spatial information of original spectral bands effectively. But
spectral distortion is still a problem in fusion result. In the
future, our work is focused on establishing a more flexible
fusion rule for information displacement.
1298