nore 
1 
sntal 
iphic 
user 
'e to 
ty in 
ning 
ivize 
1e is 
| the 
data 
done 
and 
king 
nted 
be 
Geo- 
Vich- 
ions- 
| ZU 
und 
/Info. 
und 
1994: 
welt- 
yhical 
tional 
Vol. 
RESEARCH ON TOPOGRAPHIC CONTOUR MAP SOFTCOPY 
Ma Fei 
(Research Center for GIS, WTUSM, 39, Luoyu Road, Wuhan, 430070, China) 
Commission IV, Working Group 3 
ABSTRACT: 
This paper discusses study and experiments on topographic contour map softcopy. The main aim of this research is to 
develop a practicable software system on Windows for scanned contour map automatic recognition and then to 
generate corresponding Digital Elevation Model (DEM). In the system, raster data is organized in the form of quadtree 
structure, and vector data is manipulated on Freeman code. At first, image processing insists of two aspects, that are 
frequency domain processing and spatial domain processing. In frequency domain, fast Fourier transforms and 
wavelet transforms are used to delete noise from optical scanning, and the result helps to contour line detection. In 
spatial domain, mathematical morphology, a new complete theory on parallel dealing raster data, is used to improve 
image quality, to segment contour lines from map symbols, and then to implement features of symbols and contour 
lines. Gray-scale thinning, an efficient algorithm, makes contour thinning and backgrounds optical illumination 
submerges disappear, and then it is easy to determine threshold to make gray scale image binarilized. As follows, 
binary morphological thinning makes contour one pixel wide, and short arcs are deleted. Contour gap connecting, 
which is difficult during the whole procedure and needed in order to make raster data vectorlized, is dealt with on 
morphological transform and knowledge based deduction. Another difficult question, that is contour height determine, 
is solved by inference from morphological points and lines. At last, DEM is constructed in the form of triangulated 
irregular networks. Restore DEM data is discussed as well. Experiments on 1:50,000 topographic contour maps verify 
the efficiency of the system. 
Key Words: 
Topographic contour map; Scanned image processing; Digital Elevation Model; Mathematical Morphology; Map 
recognition 
1. INTRODUCTION 
The significance of constructing topographic database in 
a large area such as a district takes attention to those 
who work on city planning, land management and 
resource development. The first thing to construct 
topographic database is to capture enough data for 
analysis. Conventional geodesy and engineering survey 
is far from the demand of database revision on neither 
efficiency nor finance. However, our photogrammetrists 
around the world have used our own tool to change the 
dilemma. |t is remote sensing image and aerial 
photographs which contain potential information, and 
these information are often reflected on map. In general, 
the method to set topographic database is divided into 
two parts: one is extract information from. existing map, 
the other followed by the former is to update the data in 
the database which demonstrates the new situation on 
image. There are two conventional methods to change 
map information into topographic database data: one is 
to digitize the map on the digitizing table with manual, the 
other is to recognize the map by computer. Obviously, : 
the latter is more efficient than the former. Therefore, 
more and more people pay their attention to automatic 
map recognition. 
In this paper, a complete procedure for topographic 
contour map softcopy is presented. The general idea to 
make topographic contour map digitize and then to 
generate corresponding digital elevation model (DEM) is 
introduced in section Il. In more detail, scanned map 
image processing in both frequency and spatial domain 
is analyzed in section Ill, and a new automatic raster- 
vector conversion method is presented in section IV. 
DEM generation and data restoration are discussed in 
section V. At last, experiments on a new software system 
are remarked and the result is concluded in section VI. 
2. GENERAL IDEA FOR SCANNED CONTOUR 
MAP SOFTCOPY 
For a long time, toward automatic recognition contour 
map is an appealing aim to study. This study involves 
such fields as image processing, pattern recognition, 
surveying and mapping, geomorphology and database. 
The difficult is not only extracting information from 
scanned image but also setting up topological 
relationship around map and making digitized results 
meet a given precision. Therefore, the method and tool in 
the task should be selected carefully and reasonably. At 
first, image deform arouse from optical scanning must be 
analyzed to rectify and its characteristics should be fused 
into image processing. So in the first part of this study, 
mathematical mapping and Fourier analysis is needed. 
Secondly, contour lines extraction from a topographic 
map is not equal to ordinary edge detection due to 
position precision and the influence of optical shadow. 
Thus the question insists of two parts: extract watershed 
  
529 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996