2004
International Archives of the Photogrammetry, Remote Sensin
Furthermore, materials like soil and tiles present the
same phenomenon and as a result many tile rooftops
are classified as soil (Fig. 15, 16).
Figure 16. Classified image
Finally, the estimation of the classification was
achieved using 84 points leveled for each class (12
points per class). This procedure provided the error
matrix and the Kappa statistics. The accuracies
attained are presented in table 17.
ified ima Accuracy Kappa
Classified image totals (%) statistics (96)
SYNTHETIC 1 80,95 77,78
Table 17. Total results of the classification
3.2.3 Classification improvement: The traditional
techniques of processing urban areas images, due to
their high spatial frequency, when based only to the
spectral observation of the objects, do not always
provide the right result. In order to solve this
problem the procedure of the classification can be
supported, besides from some special techniques, by
further data, which can be from a topographic or
thematic map to extra layers. Also, a Digital Surface
Model (DSM) can be an important help for
separating the classes in the classified image. The
hypsometric information can be combined in the
classification, so as to achieve better results in
separating classes with similar spectral behavior.
4. CONCLUSIONS
Remote sensing, with the cóntinuously ongoing
spatial resolution of modern satellites and the
development of the processing methods for the
219
g and Spatial Information Sciences, Vol XXXV ‚Part B-YF. Istanbul 2004
corresponding data, can offer irreplaceable
cartographic products (Tsakiri a.o. 1998), like the
synthetic multispectral image with a spatial
resolution of 1 meter, which became the background
for the applications described above. Specifically,
we saw how useful can the synthetic image be,
combined with cartographic data, in visual
interpretation of urban characteristics. This could be
applied for the creation of reliable maps of several
themes, where the information, compared to a
classic map, is more and is not represented by
geometric schemes and colors, but by their actual
characteristics (texture, color, relief etc.) Also, the
classification of a synthetic image of an urban area
can have as a result thematic maps related to land
use, from which useful conclusions can be derived,
about the land use, for a period of time, showing the
tension of urban development. This way they
contribute to the procedure of making serious
decisions.
S. REFERENCES
References from Books:
Dermanis A., (1999) : Space Geodesy and
Geodynamics - G.P.S., Editions Ziti, Thessaloniki.
Karnavou E.(2000) : Introduction to Urban Planning
Notes of the Course of the Department of
Cadastre, Photogrammetry and Cartography of the
Faculty of Rural and Surveying Engineering of
Aristotle University of Thessaloniki, Editions
Department of A.U.Th., Thessaloniki.
Livieratos E., Fotiou A. (2000) : Geometric Geodesy
and Networks, Editions Ziti, Thessaloniki.
Tsakiri-Strati M. (1998) : Remote Sensing, Notes of .
the Course of the Department of Cadastre,
Photogrammetry and Cartography of the F aculty of
Rural and Surveying Engineering of Aristotle
University of Thessaloniki, Editions Department of
A.U.Th., Thessaloniki.
Tsakiri-Strati M. : Remote Sensing Applications :
Notes of the Course of the Department of Cadastre,
Photogrammetry and Cartography of the Faculty of
Rural and Surveying Engineering of Aristotle
University of Thessaloniki, Editions Department of
A.U.Th., Thessaloniki.
Tsakiri-Strati M. : Image Merging, Notes of the
Postgraduate Studies Program of Geoinformatics of
the Faculty of Rural and Surveying Engineering of
A.U.Th., Thessaloniki.
References from Other Literature:
Papagiannopoulos A. (1982) : History of
Thessaloniki, Editions Rekos & Co., Thessaloniki.
Papagiannopoulos A. (1983) : Monuments of
Thessaloniki, Editions Rekos & Co., Thessaloniki.
Haala N., Walter V. (1999) : Automatic
Classification of Urban Environments for Database
Revision Using Lidar and Color Aerial Imagery,
