PHOTOGRAMMETRIC ENGINEERING 
9 
to reduce the 10-micron residual error level 
significantly, thus giving added impetus to 
the demand for medium and low-priced 
evaluation systems. 
The last paper, by Mr. Carbonell, supports 
the foregoing statement. The author dis 
cusses numerical applications at the Institute 
Géographique National at Paris, France. The 
amount of aerotriangulation work with the 
S. O. M. stereocomparator has increased 
greatly, thus reducing the workload on the 
Stereotopograph Poivillier BP. The two 
methods are compared in terms of accuracy 
and economy. 
The main advantage of the classic restitu 
tion approach is seen in the better identifica 
tion of unmarked points and the associated 
greater speed and quality of the pointing 
operation. 
The numerical approach, based on coordi 
nate measurements, provides for higher preci 
sion in the formation of the model and 
roughly 50% increase in the output. IGN has, 
therefore, concluded that the numerical ap 
proach has some advantages over the ana 
logue reduction method, including some re 
duction in cost. 
There is no appreciable difference between 
the two methods in the overall accuracy of 
aerial triangulation with small scale photo 
graphs. The accuracy is typically on the 
order of +20 microns for internal precision, 
and about ± 60/x for absolute accuracy (in the 
plane of the photography), for a block of 
224 photographs, arranged in 8 longitudinal 
and 3 transverse strips. 
The accuracy levels obtained in praxis, as 
mentioned in the last two papers, appear to 
be large compared with the results obtained 
by numerical photogrammetry in the testing 
and evaluation of components and the ex 
ecution of specially designed tests in aerial 
triangulation. The disagreement, however, is 
not a real one. Results with an accuracy of 
±2 to ±3 microns are obtainable not only in 
laboratory-type experiments, but in full- 
scale triangulation as well. This point was 
proven again recently by the results of an 
analytical block triangulation (9 photo 
graphs) executed by the Coast and Geodetic 
Survey, using the Wild RC-7 plate camera, 
for the study of earth crustal movements in 
Utah. The residual error for targeted points 
after adjustment was ±2.3^. 
There can hardly be any doubt that the 
present generation of precision cameras per 
forms very close to the principle of a true 
central perspective. If the potential of 
analytical photogrammetry is then used to 
absorb remaining perturbations, a high de 
gree of accuracy can be obtained in the de 
termination of directions, provided that the 
imagery is ‘hard’ and well defined. 
In production-type operations, however, 
economical consideration may very well pre 
vent one from taking all the necessary pre 
cautions. 
Unfortunately, the photogrammetric meas 
uring method is subject to a fast degradation 
in accuracy if any relaxation of the data 
acquisition or data evaluation methods are 
tolerated or become necessary for economical 
reasons. 
It is understandable, therefore, that there 
is an increasing interest in the economical 
aspects of numerical photogrammetry which 
is especially tailored to the support of medium 
to low-accuracy work. 
The aforementioned papers give evidence 
that the general photogrammetric com 
munity is fully aware of this possibility. 
Economical, easily-operated comparators are 
available for these applications, as evidenced 
by the presented papers. Aside from this area 
of application, there is also the ever-increasing 
need for more precise comparators in the 
realm of high-precision photogrammetry. 
The age of satellite photogrammetry has 
arrived. The promising prospect of geodetic 
position determination by means of photo 
grammetric operations, including auxiliary 
sensors, is not far benind. As additional 
documentation of successful experiments 
with the new methods becomes available, 
the hope is expressed that industry will re 
spond to the challenge by providing the neces 
sary measuring instrumentation.