Abstract
An experimental complete system of computer-aided contour line production
Framework of the experiment - The Topographical Work Division of the Regional Development and Town Planning Depart-
ment (DAFU/DTT) has started to draw contour lines interpolated by computer on its large-scale maps obtained by digital
photogrammetric acquisition. The DTT's search concerning an automatic method was motivated essentially by two fac-
tors : -a) Possibility of integrating the method in a planimetry acquisition system, -b) overall low-cost of system.
After studying various programs, either existing or in the process of development, the DTT chose, for a number
of reasons, to use the resources of the Atomic Energy Commission (CEA) ; the Minerology Department of this body had
studied, for several years past, the problems of adjusting surface-areas to sample values. The CEA possesses very
powerful data processing facilities, and the Minerology Department engineers are qualified System Analysts. M.Coulomb,
who is in charge of this department, has developed with his team, the processing program.
The purpose of this paper is not to define the mathematical methods used. Other papers may be written, compa-
ring the different methods, and the results given by each one. It now seems to be worthwhile, after one year of trial and
error and experimenting on the problem of the "Operational" use of data stemming from an altimetric diagram plot, to
outline the main problems encountered.
From the very start of the method, an endeavour was made to define a specification for the operator who was
performing the acquisition. Note at the outset that operators are in general plotters, i.e. their usual work consists in
following contours in a model, to achieve a rough draft, which is interpreted as and when it is plotted on the X Y recor-
der. As a result it has been necessary to ask the operators to be also topographers, which naturally causes a few diffi-
culties to arise.
The specifications relative to the processing program concerned : -a minimum density of points to allow the con-
tour lines to be calculated ; -the possibility of defining "forbidden' zones and "mandatory'' zones ; -instructions reques-
ting the operator to go outside the working area by a margin of altimetric points over 120 m, which would guarantee later
on the link-up in the event of an extension being drawn ; -practical details concerning the handling of the equipment and
the codes ; -as to the manner of arranging the points, it was recommended to avoid a systematic sweep which would only
permit good results with a very high density. On the other hand, it has been thought advisable to carefully mark the geo-
morphological lines.
First Results - The test sheets were very good, with a density of 5 to. 6 points per hectare in a 1:5 000 scale model. Com-
paring these results with those made with a stereoplotting performed in accordance with conventional processes, it was
found that the contour lines were just as significant in one case as in the other.
However a few blemishes due to the use of the automatic method have been brought to light by the usual utilizers
of maps, through the very shape of the contour line obtained. It does in fact seem that these lines need here and there a
few retouches ; as follows :
1) Smoothing : the curves are not smoothed - only the sharp edges are rounded. The topographer is loath in fact to intro-
duce deformations solely for the "beauty' of the result.
2) Retouching : This is required since it is a particular-feature of automatic interpolation that, in defining the terrain
which best matches the sample, it does not "think" to suppress a closed contour of short length at the bottom of a dip.
or at the top of a rise. This retouching tends to be required in the shallow thalwegs, and in very flat zones.
3) Interference with the planimetry - It is not possible to prevent the contour lines crossing paths and houses ; one avoids
drawing contours on important roads (mainly in cases of cuts and fills), waterways or rivers, etc., by delimiting these
zones during the acquisition by special contours called "forbidden", as mentioned above.
The experiments subsequently performed on a number of work sites gave very unequal results. This was enti-
rely due to acquisition problems, taking into account the particular types of terrain. The points, acquired too quickly
and systematically, included numerous errors, and the operator did not arrange them judiciously most of the time. The
density was very variable from one operator to the next for a same site. For nearly a year, the operators underwent a
veritable training course, including an explanation of the manner in which the downstream program processed the data,
and the checks it was able to make, etc.
Certain types of terrain were eliminated from the acquisition these consisted of very flat zones where the contour
lines at an equidistance of'2 metres trace out a splendid labyrinth, as though they were trying to find their way. A deci-
sion has now been reached to endow these zones with a pattern of points which, identified by a particular code, are resti-
tuted by their height. In this way one avoids drawing in these zones contour lines of little meaning.
Finally, a very detailed specification has been drafted by the D. T. T. for operators from whom, in the last
resort, an intelligent work is required, and which demands in any case the display to topographical qualities.
Conclusion - The establishing of contour lines by data processing procedures involves colossal programs of control and
calculation, very powerful computers, and interpolation methods about which mathematicians are ready to quarrel to gain
an accuracy of a few centimetres on the terrain. But this giant with feet of clay rests in actual fact on the quality of the
acquisition, which the fatigue of an operator may render illusory. We have created interactive control and correction
programs, using an IBM 2250 console, but finally it remains unthinkable to have to check every point. This program has
therefore served to detect experimentally the types of error and has proved very useful in drafting the definitive specifi-
cation.
In the current state of the production, it is still necessary, under certain circumstances, to make a few era-
sures or pencil corrections to the definitive plans. For the future we would like to formulate two wishes :
1. Improvement in the acquisition by the use of operators endowed with the necessary qualities.
2. Acceptance by the user of a new type of product, which may sometimes appear fairly rough, but which is of very low
cost without, at the same time, requiring unnecessary retouches to be made to the sheets.
- 66 -
MOL
pre
col
ces
anc
aut
int
phi
anc
the
cor
tog
anc
re]
ex]
ac
tri
col
ex
pr‘
ott
us:
of
ter
the
cu
