PHOTOGRAMMETRIC ENGINEERING 
5 
as an analytical plotter (e.g., the APC built 
by O. M. I. and Bendix after ideas from 
U. V. Helava). There is also a trend toward 
automation of the measuring process by 
applying image correlation techniques. These 
techniques require comparison circuitry for 
which the on-line computer can be of use. 
This entire group of developments and cor 
responding considerations are discussed un 
der the title: “Automatic Photogrammet- 
ric Instruments,” by Mr. R. D. Esten, 
GIMRADA. 
Consequently, this presentation is con 
cerned exclusively with coordinate measur 
ing comparators such as have been used by 
photogrammetry from its very beginning. 
Considering the enhancement of the over 
all accuracy of the photogrammetric method 
as the main purpose of numerical photo 
grammetry, the basic requirement for the 
image coordinate measuring process is a de 
mand for a significant increase in accuracy. 
Our present analogue-type restitution equip 
ment cannot adequately fulfill this require 
ment. 
Literally hundreds of numerical plate re 
ductions, each having up to eight hundred 
(800) star images, have been performed dur 
ing the past decade in the process of develop 
ing a precision photogrammetric method for 
geodetic satellite triangulations. Numerous 
other photogrammetric records, taken over a 
targeted test area, have also been evaluated. 
These records, all glass plates, were meas 
ured with a proven coordinate measuring 
accuracy of ± 1.0 to +1.5 microns. The mean 
errors of unit weight, after the corresponding 
photogrammetric triangulation procedures, 
were typically between ±2.0 and ±3.0 
microns. Further sophistication in the mathe 
matical models is possible and corresponding 
test evaluations indicate a reduction of the 
mean error of unit weight after adjustment to 
±1.5 to ±2.5 microns for records taken with 
cameras having focal lengths between 100 and 
500 mm, respectively. The domain of preci 
sion computational photogrammetry is, there 
fore, characterized by a percentage accuracy 
for an individual ray of from 1 part in 70,000 
to 1 part in 200,000. 
In light of these facts, it is essential that 
the overall accuracy of the coordinate meas 
uring procedure in precision analytical photo 
grammetry is at least ± 1 micron. 
The present generation of comparators 
offered by industry for numerical photo 
grammetry and instrumental component re 
search leaves much to be desired. 
There is hardly a comparator on the market 
which features sufficient magnification for the 
pointing operation or sufficient positive- 
identification resolution to rescue the elusive 
micron from the uncertainties of probability. 
The one-micron, least count read-out, 
which only lately has become generally avail 
able, is inadequate for suppressing the read 
out noise to a negligible level. 
Systematic errors caused by periodic and 
secular scale errors, by curvature and weave 
of the ways, and by lack of perpendicularity 
of the motions very often significantly exceed 
the micron level. Proper calibration is not 
always possible because of failure of the 
comparator to reproduce its actions within a 
fraction of a micron. 
Abbe’s comparator principle is more often 
violated than adhered to, and the outdated, 
classic, parallax stereo-comparator design is 
still with us. Research and development in 
precision photogrammetry in particular, 
and the execution of high-precision image 
coordinate measurements generally, are 
hampered by the unavailability of a high 
precision calibration grid, which is needed for 
economical, periodical calibration checks. If 
industry could produce a grid which is 
accurate to at least ±0.5 micron, together 
with a calibration certificate, the photogram 
metric community would have a standard 
which is suitable not only for calibrating in 
dividual comparators, but for comparing re 
sults in precision numerical photogrammetry 
between various photogrammetric centers of 
interest. 
Generally speaking, the field of precision 
comparators has not attracted enough atten 
tion from industry, when compared to the 
progress made in the development of preci 
sion sensors and numerical data handling 
capacity. 
The field of stereo or multi-plate compara 
tors, especially, has since the last Congress, 
with few exceptions, advanced only by the 
incorporation of minor modifications, with 
out any real effort to develop an instrument 
capable of shouldering the requirements of