wm
mW
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
6. CHANGE DETECTION AND MAP REVISION
All of the results of the object extraction methods converted to
vector layers. Images and aerial photographs together with
extracted vector layers and old maps imported to a geodatabase
in ArcGIS. Images saved in a raster catalog while vector layers
saved in feature datasets. Both raster catalogs and feature
datasets assigned the same coordinate system. All of the images
and vector layers added in an ArcMap project to track changes.
Two methods of change detection were used, comparison of the
old and new extracted maps and comparison of the old maps
with the recent images and aerial photographs. Images overlaid
with aerial photographs and old digital maps. We examined
those parts of the image where changes relative to the old maps
had been occurred. Several kinds of changes were evident, some
existing buildings had been vanished and some new ones had
been constructed. In the city marginal areas, agricultural lands
had been withdrawn in favor of man-made constructions. All of
these changes manually digitized and saved in a dataset (named
“Changes”) in the project geodatabase. In the second part of
change detection process, the old maps overlaid with the new
extracted maps. In this stage, different GIS methods were used
to detect areas where changes had been occurred. Changed areas
in the new maps were selected and exported as new layers and
saved in the geodatabase (in the “Changes” dataset). At last,
change layers merged together and converted to one layer. With
the previous mentioned method, the change layer overlaid on
the image to find those parts where editions were necessary.
The final changes layer then overlaid on the old maps layer.
Changed parts of the old maps and vanished features detected
and newly constructed features added in an edition procedure. A
new dataset called “UpdatedJLayers” created and each of the
revised layers of the old maps imported to it. Labels and
attributes of the newly added features inserted to each layer’s
attribute table. All of the revised layers then added to map
sessions according to the old maps indices to create new
updated maps of the project.
7. CONCLUSION
In this study, we utilized different image fusion and object
extraction methods to derive maximum information from high
resolution satellite images and explained the process of using
the extracted information for map revision. Comparison of the
image fusion methods discussed in this study, multiplicative and
wavelet methods were better in keeping spectral information,
hence were good in terms of preserving color properties of the
objects. IHS and PCA methods were better in keeping spatial
details of the objects; however, IHS method was good enough
in keeping spectral information as well. In sum, we concluded
that IHS method is better owing to its fairly keeping both
spectral and spatial information. Ikonos and QB MS images and
the pan and pan-sharpened product of the Ikonos were used to
extract different object classes. Multispectral images mainly
employed in unsupervised and supervised classification while
pan-sharpened product was a major help to extract smaller
objects. Pan-sharpened image produced by PCA and IHS
methods were very useful for detection of the boundaries of the
objects in visual interpretation method. Pan-sharpened image
produced by multiplicative and wavelet fusion methods utilized
in automatic classification procedures due to their better
performance in keeping spectral information. Results from
automatic clustering methods and fuzzy classification needed
more manual editing and visual inspection after vectorziation,
even though they cost less time comparing to visual
interpretation. Vectors extracted from classification methods
were used for the revision procedure only after visual inspection
and edition.
REFERENCES
Croitoru, A., Tao, V., Hu, Y. and Xu, J., 2004. The Rational
Function Model: A Unified 2D And 3D Spatial Data Generation
Scheme, ASPRS Annual Conference Proceedings, May 2004,
Denver, Colorado.
Di, K., Ma, R. and Li, R. X., 2003. Rational Functions and
Potential for Rigorous Sensor Model Recovery.
Photogrammetric Engineering & Remote Sensing, 69 (1), pp.
33-41.
Hu, Y., Tao, V. and Croitoru, A., 2004. Understanding The
Rational Function Model: Methods And Applications, DVD
Proc. of the XXth International Society for Photogrammetry and
Remote Sensing (ISPRS) Congress, Istanbul, (2004).
Leica Geosystems, 2003, Erdas Field Guide, Seventh Edition,
GIS and Mapping, LLC Atlanta, Georgia, p. 698.
McCall, M. K., 2003. Seeking good governance in
participatory-GIS: a review of processes and governance
dimensions in applying GIS to participatory spatial planning.
Habitat International, 27, pp. 549-573.
Ng, W.Y., Dorado, A., Yeung, D. S., Pedrycz, W. and Izquierdo,
E., 2007. Image classification with the use of radial basis
function neural networks and the minimization of the localized
generalization error. Pattern Recognition, 40, pp. 19 - 32.
Patynen, V., 1998. Digital orthophotos and map revision in
National Land Survey of Finland, ISPRS Commission IV on
GIS-Between Visions and Applications, Stuttgart, Germany,
1998,1APRS, Vol. 32, Part 4, pp. 463-466.
Sahin, H., Buyuksalih, G., Akcin, H., Topan, H., Karakis, S.
and Marangoz, A. M., 2004. Information Content Analysis of
KVR-1000 Ortho-Image Based on the Available Topographic
Maps in The GIS Environment, EARSEL Workshop on Remote
Sensing for Developing Countries, Cairo, 2004, pp. 1-15.
Zhou, J., Bischof, W. F. and Caelli, T., 2005. Human-computer
interaction in map revision systems, CD Proceedings of the II th
International Conference on Human-Computer Interaction
(HCIF05), Volume 3, Lawrence Erlbaum Associates, Inc (ISBN
0-8058-5807-5), Las Vegas, Nevada, 2005.
Yi-jin, C. and Xiao-wen, L., 2004. Cartographic Semantics and
Remote Sensing Data for Map Revision, IEEE 2004, pp. 2819-
2822.
1354