15
Fused image
edule
LYSIS
\.T SAR image of
sand 3, 4, 5) and a
:n accurately co-
l Figures 2(a), 2(b)
sion image if the
l image, the TM
nough. Figure 3 is
The fused images
'tative analysis.
2(b) Landsat-5 TM images (band 3, 4, 5)
2(c) SPOT Pan image
Figure 2.The co-registered images
Figure 3.The fused images
3.1 Visual examination
Compared with from figure 2(b) and figure 3, urban and built-
up areas, as well as roads are clearly enhanced and almost
perfect preservation of spectral signatures is visible in the fused
images. Visual examination suggests that the fused images are
higher in spatial resolution than that of the Landsat TM images,
and have better visual effect. Due to limitations of human vision,
comparison and appreciation by visual methods does not reveal
the exact potentials of the fusion methods. Hence, a comparison
of image statistics is attempted to evaluate the results of fusion.
3.2 Quantitative analysis
In addition to visual analysis, we conduct a quantitative
analysis. We base our analysis of the experimental results on
entropy, average gradient of the fused images and correlation
coefficient (Table 1) between the fused images and the Landsat
TM images.
In Table 1, we show that the entropy and the average gradient
are bigger than the corresponding values of the TM images. The
correlation coefficient between the fused images and the
Landsat TM images is very high. From these results, we can say
that the proposed fusion method provides more detailed spatial
information, simultaneously, preserves spectral content of the
MS image. The proposed approach is effective for image
interpret and classification. Its effectiveness for urban object
extraction will be presented in the other paper.
Entropy
Average
gradient
Correlation
coefficient
5.40
9.33
1.0
TM images(band
5.32
5.90
1.0
3,4,5)
6.25
13.92
1.0
The fused
5.82
10.02
0.89
5.57
7.61
0.88
images
By the proposed
method
6.53
14.55
0.90
Table 1 Evaluation of the fused image
4. CONCLUSION
A new image fusion method of SAR, Panchromatic (Pan) and
multispectral (MS) data is proposed. The SAR texture
extraction and high pass details of Pan image modulated with
the SAR texture are reasonable. The proposed method can
improve spatial resolution and simultaneously preserve spectral
content of the MS image, and is effective for image interpret
and classification. For future work, the authors would be
looking at evolving to a more autonomous fusion system.
REFERENCES
Alparone L., Baronti S., Garzelli A., et al,2004.Landsat ETM+
and SARp Image Fusion Based on Generalized Intensity
Modulation. IEEE Trans. On Geosci. RemoteSensing,
vol .42, pp. 2832 - 2839.
Forster B.C., 1985. An examination of some problems and
solutions in monitoring urban areas from satellite platforms.
International Journal of Remote Sensing, Vol.6, pp 139—151.
Hepner G.F., Houshmand B., 1998. Kulikov, etal, Investigation
of the Integration of AVIRIS and IFSAR for urban analysis.
Photogrammetric Engineering and Remote Sensing, vol.64,
pp813- 820.
Ranchin T. and Wald L., 2000.Fusion of High Spatial and
Spectral Resolution images: The ARSIS Concept and Its
Implementation. Photogrammetric Engineering and Remote
Sensing, vol.66,pp.49-61.
Pohl C. and Genderen Van J. L., 1998. Multisensor image
fusion in remote sensing: concepts, methods and applications.
International Journal of Remote Sensing, vol. 19,pp.823-854.
