bul 2004 
ge pixels 
closest 1 
int of the 
polygon. 
f nucleus 
we in the 
are taken 
ral pixel, 
natrix are 
d in this 
yrs: After 
)0 000 to 
pological 
contours 
it. for the 
netrically 
asing the 
n points 
make an 
| of sharp 
ralization 
  
1 contours 
he number 
y the user 
f the map 
lines can 
ographical 
e speed of 
important 
ions to be 
ecially in 
bor curves 
his choice 
  
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004 
2.4. Deleting the Vectors According to the Length 
This process is designed for saving time. While editing the map, 
the map and the contours have to be superimposed. To realize 
this process the CPU needs a considerable time, for example: an 
Intel Pentium 4 2.4 Ghz processor with 512 MB RAM can run 
this process for 2 million nodes in 3 seconds. This time delay 
results in loss of concentration for the user. No graphic 
accelerates (DirectDraw, Direct3D, OpenGL , etc.) have been 
used for this step of software. But all the imaging process is 
done by GDI. 
For this step, the user defined contours under a predefined 
length (in pixel value) are automatically deleted. On the other 
hand, the small cartographic objects which have the same 
threshold range with training samples are deleted automatically 
too. Then, the editing time and user mistakes are decreased. 
2.5, Editing the User Marked Contours 
Due to the scanning or deformation of the map, broken or 
wrong contours can occur in the elevation lines after automatic 
elevation line recognition process. 
The steps below are practiced to correct these mistaken lines: 
e Deleting the vectors that occur while digitizing the 
cartographic objects that were not elevation lines as 
elevation lines 
e Drawing the lines that were not drawn or that were 
deleted because of wrong formation 
e Changing the places of broken points of existing 
elevation lines 
e Deleting the broken points of existing elevation lines 
for generalization 
e Separating the wrongly joined elevation lines 
e Joining the broken elevation lines 
2.6. Determining the Elevation Values for Edited 
Contours 
The user chooses the elevation lines one by one and gives their 
values. This step can be considered not to be a lengthy process. 
When the broken lines are joined, the number of new lines is 
very few. For this reason, this step takes 30 minutes to 1 hour 
according to the capability of the user. 
A satisfactory result can only be obtained by giving Z value to 
the lines where the inclination changes. In the developed 
software, there is a command to delete the lines which are not 
given elevation values. By this method, the rest of the lines can 
be deleted at once. 
463 
3. SOFTWARE OUTPUTS 
  
Figure 2. 3D elevation model 
of results (Model Made with © ESRI ArcInfo) 
Broken contours are separated according to their XYZ values in 
ASCII file method, which is a method supported by most of the 
CAD, GIS and other engineering software systems (Figure 2). 
Thus, the contours can be saved as xyz extension files. This 
method makes it possible to get output from overloaded files on 
the hard disk in accommodation with every program. 
4. CONCLUSION 
The developed algorithm aimed to accelerate the digitizing 
process of 1:25 000 scaled maps semi automatically; 
independent from the map’s topological structure. The study 
has achieved this aim. 
Another factor that affects the method of the study is the 
resolution of scanned image. This algorithm is developed 
considering the thickness of the contours as 1 or 2 pixels.’ 
Another fact that the image quality is that, the image should be 
scanned in RGB. Otherwise, it is difficult for the algorithm to 
differentiate the contours from other cartographic objects. This 
difficulty increases the time spent for designing the lines, which 
prevents the study from reaching its goal. 
The most important difference of this algorithm is its contours- 
oriented method. The main subject of this study is recognition 
of the contours from their center points. The 1:25 000 scaled 
maps have contours of 10 m. interval. So as not to reduce the 
elevation sensitivity, the contours are recognized from their 
middle points. However, many vector algorithms realize this 
process by converting the borders of the pixels that form the 
lines to line pieces. This method increases the time spent for 
designing the contours. 
In this algorithm, the time spent for semi-automatic recognizing 
of plain and rough areas on a map is about 4 hours to 1 day. For 
extremely rough areas, this process takes 1 week to 10 days. 
With classical methods, this step takes 3 days to 1 week for 
plain and rough areas and 1 week to 1 month for extremely 
rough areas. 
The developed algorithm works with a large amount of data. 
The information gained from this data can be much more 
complex, according to the roughness of the area or the starting 
parameter that the user chooses. Therefore, this software has 
been developed in C++ language. Since the C++ language is a