Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
51 
New approach to semi-automatically generate digital elevation data 
by using a vidicon camera 
C.C.Lin, A.J.Chen & D.C.Chem 
Center for Space and Remote Sensing Research, National Central University, Chung-li, Taiwan 
ABSTRACT: A systematic procedure is proposed for semi-automatic digitization of topographic maps by using an 
interactive image processing system, including a vidicor camera. The required input is line-drawn contour map. 
A contour-line image is then generated by the vidicon camera. To produce a clean contour-line image with one 
pixel in width, standard preprocessing techiniques are adopted, such as image enhancement, noise-removal, line 
thinning, and editing etc. The resels and their corresponding boundaries are labeled automatically. Isolated 
peaks, flat ridges or troughes, are, then, added afterwards. A look-up table is manually generated, 
associating the elevation data with those points, lines or closed boundaries. The elevation value at a given 
pixel within a resel is weighted by the two smallest "effective distances" from this point to all of its 
neighbors, such as isolated points, lines or boundaries. The effective distance, defined here, represents a 
shortest distance between a point and a curve with the connecting path lying within a resel, i.e., the 
intersection with the contour lines are not permitted. A,fast computing algorithm is developed to calculate 
the effective distances of all the pixels within a resel. The construction of a digital elevation model is 
easily achieved. The main advantages of our approach are two-folds. Firstly, the variation of the elevation 
values within a resel is well behaved, regardless the complexity of a resel, even for elongated kindly shape, 
etc. Secondly, the ridges and troughes are automatically generated and clearly repersented. A simulated 
results will be presented to illustrate the processes and demonstrate its advantages. 
1. Introduction: 
Topographic data are nowadays most commonly 
obtained through the technique of comparing optically 
photographed stereo pairs of imagery. Large amount 
of manpower is needed for applying this method. 
However, owing to its great importance, contour maps 
are prepared with various degrees of accuracy for 
most parts of the world with this or other less 
efficient methods. 
Satellite images, so far, are not suitable for 
extracting precise topographic information(Mckeown, 
1984). Yet for many geological, hydrological, 
forestry or enviromental applications, it is 
necessary to have the digital elevation model (DEM) 
(Ragan 1975, 1978). The DEM, thus, also constitutes 
an indispensable layer for constructing geographic 
information system(GIS) (Ragan et al., 1976; Colwell, 
1983; Faintich, 1984). 
Interpolation methods have been used to generate 
DEM (Tempfli, 1977; Grimson, 1981; Sincar et 
al,1984). One of the approaches is that, based on 
the given elevation values of some points the 
elevation values of other points in a region are 
interpolated (Rioley, 1981). Softwares using this 
approach are available commercially (SACM, 1981). 
Recently, raster scanning system, such as vidicon 
camera, becomes more widely available. With the aid 
of such system, we can use the contour line map, in 
raster format, as the basis of the interpolation. In 
a recent article, Sircar and Ragan (1985) described 
such a interpolation process, with a method of 
steepest decending and 4 reference lines. And it is 
adaptable to the small users who need to obtain a 
reasonable accurate DEM in a short period of time. 
This process, though efficient, cannot spread the 
elevation information uniformly in certain types of 
terrains, e.g., in an elongated kidney-shaped trough, 
the. elevation information in one end of the trough 
may not be conveyed to the other end. Nevertheless, 
such types of terrains are frequently encountered in 
a mountainous area where DEM is particularly useful. 
An algorithm that is capable of spreading the 
elevation information uniformly in any type of 
terrain is presented in this paper. The algorithm is 
also suitable for small user to generate DEM quickly. 
In section 2.-, prepossessing of the contour line 
image is discussed. In section 37 , effective 
distance within a resel is defined. In Sesction 4., 
a fast algorithm is devised for finding the 
interpolated elevation values. In section 5., an 
application example is presented. Finally, 
concluding discussions are offered in section 6V. 
2. Digitized Contour Line Image and Preprocessing 
The image produced by a vidicon camera from a 
contour line map, e.g., the line-drawn map shown in 
Fig.l, usually has some defects. Hence some 
preprocessings are always necessary. The purpose of 
Fig. 1. The line-drawn map copied from topomap.