The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
1284
resolution images with the high resolution images if in a step-
by-step approach the input files and their spectral structures are
controlled and prepared for the final merging. In this way, the
results are much more satisfactory than the ones obtained
through the Direct Merge method. For example, several linear
and point features (wood fences, trees, etc.) that are difficult to
distinguish in the Landsat images are visible in these merged
images.
2.3 Merge with Resampled Landsat Imagery
In this method, at the first step multispectral Landsat 7 ETM
imagery was resample to higher resolution imageries (20, 10, 5
meters resolutions) step by step. At the second step, these three
images were enhanced with filtering. Finally, panchromatic
aerial imagery was merged with these this resampled Landsat
imagery again with three different merging techniques. The
merged images are shown in Figure 6 and also a part of imagery
is zoomed in and shown in Figure 7.
Figure 6. Merged imageries (a: Brovey Transform, b: IHS, c:
Principal Component, d: Resampled Landsat multispectral)
Figure 7. Merged imageries (a: Brovey Transform, b: IHS, c:
Principal Component, d: Resampled Landsat multispectral)
In this method, Brovey Transform gives the more satisfactory
result by creating a good spectral reflectance. This method
decreases the color spot effect in the direct merge method and
creates a more homogenous merged imagery.
3. CONCLUSIONS
Merging methods for utilising both the high resolution
panchromatic and the multispectral images in a combined
manner is one way of improving the methods for many remote
sensing applications such as change detection, classification, etc.
The aim of the resolution merge is to achieve a maximal spatial
detail augmentation and a minimal color distortion.
The results of this study demonstrate that for the merge of aerial
photographs and Landsat ETM imagery, Principal Component
and Brovey Transform techniques achieve significantly better
results than IHS. The main advantage of the images generated
by the Principal Component and Brovey Transform techniques
is that they have a close spatial characteristics with the aerial
photograph. This could make the photo interpretation easier.
Furthermore, these images have an aspect of softly colored
aerial images, in which the color tones have been obtained from
the Landsat ETM ones. The merged images are more similar to
the Landsat ETM multispectral images, but with better spatial
feature details derived from aerial photographs.
Another conclusion is that merge with resampled Landsat
imagery generates a better spectral merged imagery compared
with the other methods.
IHS technique doesn’t work well for imageries having such a
big difference of spatial resolution. For this technique, merge
with resolution merged Landsat imagery gives the best results.
A lack of this technique is that since this technique creates only
three spectral bands, it is not preferred for some remote sensing
applications such as classification.
Brovey Transform technique creates very sharp merged images
but with some spectral problems. In direct merging, some color
spots happen in the merged imagery. Merge with resampled
Landsat imagery decreases this effect.
In all of these techniques, Principal Component seems to work
best for such imageries. It creates imagery with low spectral
problems. Only the sharpness is less if compared with the
Brovey transform. But this problem can be decreased by
merging with resolution merged Landsat imagery.
As a result of this study, it is advised that for the applications
needing good spectral reflectancy; Principal Component
technique and Merge with Resolution Merged Landsat Imagery
method should be used and for the applications needing good
spatial information; Brovey Transform technique and Merge
with Resampled Landsat Imagery method should be used.
REFERENCES
Bretschneider, T., Kao, Odej.2004, Image fusion in remote
sensing, Technical University of Clausthal, Germany,
http ://www. ntu. edu. sg/home/astimo/Publications/Documents/O
SEE2000.pdf
Carvalho, J., Delgado-Garcia J., Soares, A., Caetano, H.,2006,
Stochastic simulation with reference images for merging