ibul 2004
lencies, a
) area.
jects are
ique can
object is
extended
- top). If
trimmed
ses small
another
ll. edges,
lowever,
rocessing
dges are
posed to
S in size
rlaps are
' can be
ntered by
mensions
normally
] even 1f
se object
ramming
ein the
file) and
> format.
rmats of
'ersity of
; see IfK
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
BUHOM uses a configuration file for user parameters. In this
file (buhom.cfg) the following parameters are entered by user:
e Search radius for identical points
e Angular threshold value for right angle adjustment,
for recognition of insignificant points, and for
parallelism to street.
e Threshold value for small area
e Maximum distance from building to street (/ in figure
3)
e Are small objects deleted or not? (y/n)
e Are identical points eliminated? (y/n)
e Self-cutting test? (y/n)
e Test on insignificant points (y/n)
e Insignificant points are to be eliminated or aligned?
(y/n)
e Right angle adjustment (y/n)
After processing a report file is created, in which some useful
information about last program session is to find.
BUHOM do not correct gaps and overlaps that can occur after
right angle adjustment, because such inconsistencies are
removed in the following generalization process with
CHANGE. If generalization does not follow the enhancement,
these inconsistencies can be corrected with another program
(GQE), which is developed by Bildirici (2000). GQE
(Geometric Quality Enhancer) is also developed with
FORTRAN 90 programming language with the aim of
correcting gaps and overlaps between buildings (figure 6). In
this program the approach for topologic inconsistencies
discussed in 2.4 is realized. In the future the authors are
planning to integrate the functionality of GQE into BUHOM.
In order to give an idea about the functionality of BUHOM, an
application is given in figures 7 and 8, in which original objects
and processed objects can be seen. In this application all tests
were performed, but no parallelism to street. In figure 8, a
building with a complex outline is given. This shows the
functionality of the program visually.
4. CONCLUSIONS
Vector data is prone to geometrical inconsistencies, which
should be examined and corrected before GIS applications and
generalization. This is called line cleaning, which is integrated
in most GIS software. Line cleaning tools, in general, are
developed to process lines objects rather than polygons. Since
generalization requires error free data, especially for buildings,
a preprocessing step is needed. Therefore the program BUHOM
was developed. In addition to common line cleaning tasks, it is
capable of doing right angle adjustment and reconstructing
parallelism to streets. It has been tested at Istanbul Technical
University, Istanbul, and at the Institute for Cartography and
Geoinformatics, Hannover (Bildirici, 2000; Bildirici, 2003).
BUHOM can be used both as a preprocessing tool before
generalization and as an independent tool for
geometric/topologic improvement.
ACKNOWLEDGEMENT
The authors thank Prof Dr. Monika Sester for her
encouragement and valuable suggestions.
237
REFERENCES
Bildirici, 1.O., 2000. Generalization of buildings and roads in
scale range 1:1000-1:25000. Dissertation, Graduate School of
Istanbul Technical University, Istanbul, Turkey. (in Turkish).
Bildirici, I.O., 2004. Building and road generalization with the
CHANGE generalization software on Turkish topographic base
map data. Cartography and Geographic Information Science,
31(1), pp.43-51.
Kruse, I., Powitz, B.M., 1990. Die netzwerkartige Datenbank
des IfK, Institut für Kartographie, Hannover (unpublished
document).
Original data
— Processed data
Figure 7. Original and processed data
(Original scale 1:1000, reduced to 1:2 000)
Original data
Processed data
Figure 8. One of the processed buildings (enlarged)
