|
|
|
Poster Session 1
|
Automated Cartographic Line Tracking
Experiments
A portion of a scanned contour map sheet with scale of 1:50,000 was used in this experiment (Figure 4). It covers an area in the northern part of Thailand. The chosen part is compromised of several types of terrain data including rivers, mountains, ridges, and etc. The minimum elevation is 480 meters while the maximum elevation is 941 meters Contour lines are very dense. The original line thickness falls between 5 to 11 pixels. For the sake of simplicity, all contour annotations and other text appeared in the original scanned image were removed before this tracking algorithm performed. It can be noticed that some features in the original image are poorly defined. In addition there are some breaks in lines appearing on the original image. The grayscale scanned image was threshold to black and white image to differentiated the object pixels from the background pixels.
The output image after thinning and gap filling could be accepted in a certain regions. Some breaks in lines still appeared in the output. In general, the thinning operators worked very well on a contour map. All lines become single pixel wide features. The node marking algorithm then was employed for finding all types of nodes in the thinned image. Each thinned line then was followed on a line by line basis from its starting node until its ending nodes was reached. Note that there was no action taken on the isolated nodes.
In order to apply this contour data for use in constructing a digital elevation model or DEM, the elevation attribute data for these contour lines were noted. The individual contour lines were tagged with an elevation attribute. These XYZ contour line points could be used in several ways to build a DEM. Taken as random spot heights they can be processed into a TIN. A constant shade rendering is shown in Figure 5.
Figure 5: A shaded rendering of the terrain model
Conclusions
This paper has described an automatic line tracking technique for contour imageries. It integrates two crucial steps into one suite. The thinning operators can preserve shape of the object while it does not create any artifact or holes. At the end, the algorithm generates the series of coordinates of the thinned lines between two detected endpoints. Using this approach, memory space required to store the data can be reduced while most distinct patterns are retained. Several experimental results imply that this approach offers an alternative method of automatic line tracking.
References
- Apaphant, P. and Bethel, J., 2000. In: International Archives of Photogrammetry and Remote Sensing, Amsterdam, The Netherlands Vol. XXXIII, Part B3, pp.267-273.
-
Greenlee, D. D., 1987. Raster to vector processing for scanned linework. Photogrammetric Engineering and Remote Sensing, Vol.53, No.10, pp.1383-1387.
-
Landy, M. S. and Cohen, Y., 1985. Vectorgraph coding: Efficient coding of line drawing. Computer Graphics and Image Processing, Vol. 30, pp.331-344.
-
Moore, L. R., 1971. Software for cartographic raster to vector conversion. In: International Archives of Photogrammetry and Remote Sensing, Washington, D.C., Vol. XIV, No.5, pp.335-345.
-
Peuquet D. J., An examination of techniques for remote formatting digital cartographic data/part 1: The raster to vector process. Cartographica, Vol.18, No. 1, 1981, pp.34-38.
-
Rosenfeld, A. and Kak, A. C., 1982. Digital Picture Processing(2nd edition), Academic Press, Vol. 2, pp.55-163.
|
|
|
|
|
|
|
