You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Proceedings of the Symposium on Global and Environmental Monitoring

Combined Application of Pattern Recognition Methods
in Automatic Digitization of Contour Line Maps
Jun Yang
The Digital Elevation Model (DEM) has been applicated more and more in spatial
analysis, regional planning, project designing, topographical displaying etc. Many
governmental and private institutions are engaged to gather digital elevation
information from different sources. The digitization of existing maps is a highly
efficient and economic way-, if the qualities of the maps are sufficient. But the
most used manual method of digitization cannot cover the increasing require
ments. Therefore a system has been developed to automatic structuring of con
tour line maps from scanned data. The system has been tested in the Institute
of Cartography, University of Hannover.
A contour line map usually contains three kinds of information:
1. the contour lines, many of which are drawn in regularly dashed lines,
2. the numbers showing the heights of the contour lines and
3. the special symbols describing the sudden change of the terrain,
such as cliffs and outcrops.
This system is composed of the following procedures:
Scanning and vectorization. A map is transformed into image by a scanner. The
raster data have to be converted into vector data in order to reduce the data
and accelerate the processing.
Segmentation of contour lines from texts and symbols. At first all kinds of data
are in one level. The recognitions can be carried out only succesively and me
thods used for every kind of object are different. Therefore it is neccesary to
separate the data into different object subset before classification. To distingu
ish the different objects the features which contain the characteristic informa
tions, e.g. line length, line width, line curvature and number of lines on nodes,
are extracted. Upon this feature space a training process should be effected to
get the parameters that will be used in classification. At first step the contour
lines are selected from another data set by means of Quader Classification me
thod .
Recognition of contour lines. The contour lines for example in Germany Basic
Map Series 1:5000 were drawn in dashed lines excepting the contour lines at
the 10 meters vertival interval. The different line width, line length and gap
length have been used to construct the line signatures for representing different
vertival interval. The line signatures have the defined structures.