The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B3h. Beijing 2008
Figure 4. Classified result of road
4.6 Road extraction by using run length
Directional lines are drawn in all directions from the central
pixel to the peripheral circle. These lines are shown in Figure 5.
Run length of “1” on the directional line is calculated. If many
lines that run length is the same size of the radius of the
peripheral circle are gathered in one group, this direction
correspond to road. If above line group exist at least 3, this
central pixel is recognized as road intersection.
Such procedure is carried out for each pixel by moving double
circle.
Figure 5. Directional lines
5. EXPERIMENTAL RESULTS
Extracted road intersections are shown in Figure 6. Recognized
accuracy was 34%.
Some extracted errors are shown in wide roads and dotted roads.
Supervised classification result does not satisfy for the road
intersection extraction. We cannot find out the effectiveness of
the proposed double circle. We detected roads from the object
image by using visual interpretation. Double circle algorism
was applied for the obtained image. As the result, extracted road
intersection shown in Figure 7 was obtained. Effectiveness of
proposed algorism is shown in Figure 7.
Figure 6. Extracted Road Intersections from The Classified
Image
Figure 7. Extracted Road Intersections from The Visually
Interpreted Image
6. CONCLUSIONS
New method was proposed for detection of road intersection.
This method is combined supervised land cover classification
and double circle. We could show the effectiveness of this
method. There are some problems in the classification. We will
make effort to show the effectiveness of proposed method.
ACKNOWLEDGEMENTS
The author wants to thank Hitachi Software Engineering for
providing QuickBird image.
REFERENCES
[1] Hosomura, T, 2005, House Detection from High Resolution
Satellite Image by Using Double Windows, Proc. of ACRS2005
507
