|
|
|
Poster Session 1
|
Geometric and spectral analyses of
Merged remotely sensed images
Visual Image Interpretation
As shown in Figures 4-9, it may be noted that the merged data set with SPOT XS3, SPOT PAN,
and SPOT XS1 using the band substitution method has a very similar spectral appearance what
the original SPOT XS has. The merged data set with SPOT XS3, digitized aerial photography,
and SPOT XS1 using the band substitution method has greater spatial variability, but worse
spectral details. The merged data set with SPOT XS3, digitized aerial photography, and SPOT
XS1 by the band arithmetic operation method has fine geometric as well as spectral details.
The results of the color space transformation and substitution method do not have good
improvement as another two methods have. This is due to the SPOT XS image contain an
infrared band, the transformation equations used might not be suited for.
Unsupervised Classification
In the unsupervised classification the spectral classes in a scene are found automatically. This
approach has been found suitable for generalized land-cover mapping. There are many
clustering algorithms that have been developed to determine the natural spectral classes present
in a data set. A widely used clustering algorithm is the Iterative Self-Organizing Data Analysis
Technique (ISODATA). This method is used, therefore, to classify the merging images.
Classified data is then compare with classified data of the original SPOT XS data.
The unclassified percentage and total accuracy of error matrix of the classification map for
each merging image are listed in Table 1. The merged data set with SPOT XS3, digitized aerial
photography, and SPOT XS1 by the IHS method has been found to have the lowest unclassified
percentage. It means that the spectral characteristics of this merged data set is similar to what
the original SPOT XS has. Furthermore, no one total accuracy of all merged data sets has
exceeded 20%. It shows that the greater part of merged data set has been changed, but is still
related to original one.
Table 1. Unclassified percentage and total accuracy of error
matrix of the classification map from each merging image
| Merging Method | Unclassified
percentage (%) | total accuracy (%) |
| Band substitution (SPOT PAN) | 47.30 | 8.11 |
| Band substitution (Aerial data) | 46.80 | 16.93 |
| IHS(SPOT PAN) | 1.94 | 9.19 |
| IHS(Aerial data) | 0.00 | 11.41 |
| Band operation (SPOT PAN) | 41.25 | 13.32 |
| Band operation (Aerial data) | 0.12 | 5.60 |
Summary
Three different data sets have been merged and three methods used to extract the best
information from both geometric and spectral components. As mentioned above, merging the
SPOT XS and digitized aerial photography into single data sets having spectral (green to near IR)
and spatial (1m) properties.
Although the IHS transformation does not have the natural visual results, the integrated
images of this method still have the best classified effect.
Merging the SPOT XS3, digitized aerial photography, and SPOT XS1 data by th use of the
band arithmetic operation method has improved both geometric appearance and spectral
interpretation. The resultant merged data sets also have a low unclassified percentage.
Reference
-
Chavez, P. S., Jr., 1986. Digital merging of Landsat TM and digitized NHAP data for 1:24,000
scale image mapping, Photogrammetric Engineering & Remote Sensing, 52(10), pp. 1637-1646.
- Chavez, P. S., and J. A. Bowell 1988. Comparison of the spectral information content of
Landsat Thematic Mapper and SPOT for three different sites in the Phoenix, Arizona Region,
Photogrammetric Engineering & Remote Sensing, 54(12), pp. 1699-1708.
- Ehlers, M., M. A. Jadkowski, R. R. Howard, and D. E. Brostuen 1990. Application of SPOT
data for regional growth analysis and local planning, Photogrammetric Engineering & Remote
Sensing, 56(2), pp. 175-180.
- Gonzalez, R. C. and R. E. Woods 1992. Digital image processing, Addison-Wesley.
- Grasso, D. N. 1993. Applications of the IHS color transformation for 1:24,000-scale geologic
mapping: A low cost SPOT alternative, Photogrammetric Engineering & Remote Sensing,
59(1), pp. 73-80.
- Jensen, J. R. 1996. Introductory digital image processing: A remote sensing perspective,
Prentice Hall.
- Jensen, J. R., E. W. Ramsey, J. M. Holmes, J. E. Michel, B. Savitsky, and B. A. Davis 1990.
Enviromental Sensitivity Index (ESI) Mapping for Oil Spills Using Remote Sensing and
Geographic Information System Technology, International Journal of Geographical
Information Systems, 4(2), pp. 181-201.
- Schowengerdt, R. A. 1983. Techniques for Image Processing and Classification in Remote
Sensing, Academic, New York.
- Welch, R. and Ehlers 1987. Merging Multiresolution SPOT HRV and Landsat TM Data,
Photogrammetric Engineering & Remote Sensing, 53(3), pp. 301-303.
 |
Fig 1. Aerial
Image | Fig 2. SPOT PAN
| Fig 3. SPOT XS |
|
|
|
|
Fig 4. Band substitution (PAN) |
Fig 5. Band substitution (Aerial)
|
|
|
|
|
|
Fig 6. IHS (PAN) |
Fig 7. IHS (Aerial)
|
|
|
|
|
Fig 8. Bands Operation
(PAN) |
Fig 9. Bands Operation
(Aerial)
|
|
|
|
|
|
|
|
