The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B4. Beijing 2008
object of similar reflectance properties, it could be considered
as verified, too, despite the absence of the structural indicators.
Another idea to enhance the segmentation results is to use a
priori knowledge about the typical shape of management units
to introduce additional constraints. Using the information that
the boundaries of management units usually consist of straight
line segments that are orthogonal or nearly orthogonal could
improve the results for the examples given in Figure 7 and
Figure 8. Unfortunately, these geometrical constraints could
hardly improve the segmentation result in Figure 9.
c) d)
Figure 8. a) Original image, taken from the same IKONOS
scene as Figure la; b) Results of segmentation
(smoothness for Watershed = 1); c) Results of
segmentation (smoothness for Watershed = 5);
d) white lines: segment boundaries from c), black
lines: results of edge detection.
variations of the soil properties, the homogenous regions are to
a large degree coherent with the different management units
existing in a GIS object. The segmentation algorithm is still
work in progress, but the preliminary results presented in this
paper show the potential of the algorithm for the verification
approach. Even though a complete segmentation of
management units seems to be impossible, the segmentation
algorithm enhances the automatic verification process of GIS
object. The level of segmentation that could achieved is already
an important improvement of the verification approach.
Future work comprises an improvement of the segmentation
algorithm, e.g. by introducing additional (geometrical)
constraints, and the implementation of the synthesis of the
verification results achieved for the individual segments: at this
instance, segmentation errors could be compensated.
Furthermore, a more detailed evaluation of the improvement
achieved by the segmentation is to be carried out.
ACKNOWLEDGEMENTS
This work was supported by the German Federal Agency for
Cartography and Geodesy (BKG). It was also supported by the
Early Career Grant 600380 of the University of Melbourne.
REFERENCES
Barista, 2008. Barista product information webpage,
http://www.baristasoftware.com.au (Accessed: 30. April 2008).
Brügelmann, R., Fòrstner, W., 1992. Noise estimation for color
edge extraction. In: Robust Computer Vision, Wichmann,
Karlsruhe (Germany), pp. 90-107.
Busch, A., Gerke, M., Grünreich, D., Heipke, C., Liedtke, C.-E.,
Müller, S., 2004. Automated verification of a topographic
reference dataset: system design and practical results.
International Archives of Photogrammetry & Remote Sensing
XXXV-B2, pp. 735-740.
Figure 9. RGB IKONOS image (left); grouping result with
smoothness for watershed = 1 (middle); grouping
result with smoothness for watershed = 5 (right)
4. CONCLUSIONS AND OUTLOOK
Förstner, W., 1994. A framework for low level feature
extraction. Computer Vision - ECCV '94. Proc. 5th ICCV, vol.
II, pp. 383-394.
Fuchs, C., 1998: Extraktion polymorpher Bildstrukturen und
ihre topologische und geometrische Gruppierung. PhD Thesis,
Institute of Photogrammetry, Bonn University, DGK-C 502.
Gonzalez, R., Woods, E., 2002. Digital Image Processing. 2 nd
edition, Prentice Flail, Upper Saddle River (NJ), pp. 622-624.
Grote, A., Butenuth, M., Heipke, C., 2007. Road extraction in
suburban areas based on Normalized Cuts. International
Archives of Photogrammetry, Remote Sensing and Spatial
Information Sciences, 36(3/W49A), pp. 51-56.
Helmholz, P., Gerke, M., Heipke, C., 2007. Automatic
discrimination of farmland types using IKONOS imagery.
International Archives of Photogrammetry, Remote Sensing and
Spatial Information Sciences, 36(3/W49A), pp. 81-86.
The goal of the presented segmentation algorithm was the
enhancement of an algorithm for the verification of tilled
cropland objects by splitting that object into homogenous
regions. This is necessary because the verification algorithm
does not work successfully with GIS objects containing more
than one management unit. Despite disturbances caused by
Luo, J., Guo, C., 2003. Perceptual grouping of segmented
regions in color images. Pattern Recognition 36(2003):2781-
2792.
390