Neen 
/ided 
lope 
be a 
>, 
e the 
ation 
ence 
t be 
oints 
re is 
mmit 
teps: 
of a 
sink 
ce in 
e for 
ly to 
jit or 
lid in 
crete 
with 
di 
three 
ance 
other 
lates 
eton. 
d as 
  
5.2. Generating The TIN By Individual Points 
According to the definition of Thissen polygon, the 
skeleton |. S(X^) is just the Thissen polygon 
corresponding to X in raster domain. With the algorithm 
of partial painting, it is easy to find the two points in two 
different Thissen polygon which share a same side. The 
constructed triangulation through connecting all of these 
two points is just Delaunay triangulation by definition. 
5.3. Generating The TIN In Consideration Of 
Geomorphologic Points And Lines. 
In order to describe terrain exactly, geomorphologic 
points and lines must be contained in the TIN. That 
means, geomorphologic points should be triangular 
points and geomorphologic lines must be preserved as 
triangular sides. 
5.4. DEM Data Restoration 
Ordinary topological relationship of TIN containing points, 
sides and near triangular is too space consuming to 
represent. It is necessary to compress TIN relationship 
data in the case of manipulating and analysis. The 
method of regular storing TIN is to enlarge TIN to regular 
networks in topology. As you can image, any midpoint 
polygon can be thought as constitution of hexagon(s) 
topological, only have some coincided points and lines. 
6. EXPERIMENTS AND CONCLUSION 
Based on the above procedure, a new scanned contour 
map recognition system on windows has been designed. 
  
Fig. 1. Contour Map After Automatic Recognition 
533 
Experiments with the system have been done a contour 
map at scale 1:50000, automatic raster to vector 
conversion result and part of its corresponding TIN is 
shown in Fig. 1 and Fig. 2. On experiments, it is verified 
that a topological contour map without too much complex 
can be automatically digitized within 48 hours or so on 
PC 486 without any other hardware. 
REFERENCES: 
1. Wang Zhizhuo, Principles of Photogrammetry (With 
Remote Sensing), Press of Wuhan Technical 
University of Surveying and Mapping, Wuhan, China, 
1990. 
2. Li Deren, Chen Xiaoyong, Ma Fei. Applications of 
Mathematical Morphology in automatic Cartography 
and GIS. International Archives of Photogrammetry. 
Volume 30-4. ISPRS Commission IV,1994. pp.127- 
134. 
3. Serra,J. Image Analysis and Mathematical 
Morphology. London: Academic Press, 1982. : 
4. Sternberg,S.R. Grayscale Morphology. Computer 
Vision Graphics and Image Processing, Volume 35, 
1986, pp. 333-355. 
5. Kamada,K., et.al.: Automatic Digitization of 1:25,000 
Scale Topological Maps. Volume 29-4. ISPRS 
Commission IV, 1990. pp. 86-94 
6. Li Deren, Gong Jianya. A Unified Data Structure To 
Represent Vector And Raster Data In GIS, 
International Archives of Photogrammetry and Remote 
Sensing. Volume XXIX, ISPRS, 1992, pp. 768-773. 
7. Poul Frederiksen, et.al.: A Review of Current Trends in 
Terrain Modelling. ITC Journal, No. 2 ,1985, pp. 101- 
106. 
i Ay. 
e 
m 
x Ni 5s g dos 
SE Ofen a Ë 
RE NA ; 
riva 
Ps IX 
TM 
E 
Euros 
Cs 
n 
E 
Rar PA VA Foxy OE NE 
; AN 
A A 
EIE 
bp 
n 
Fig. 2. Part Of Tin Corresponding To The Above 
Contour Map 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996