Accuracy
What has been experienced in C&GS [6] may serve as an
example for purposes of discussion of the accuracy that has
been obtained with the monocomparaior system. I am sure that
any one of you can improve on these results. For example,
the camera was an RC-7 glass plate type with a 10 centimeter
Aviogon lens. The flight altitude was 900 meters. Nine
photographs were involved in three strips of three photographs
each where the overlap was 60% in both directions. The nine
control points were signalized. Based on four of the control
points, the standard error space vector for all nine points
was 3*5 centimeters on the ground or about 0.000 04- of
the altitude. Thus the standard plate error was about 3*5
microns including all error sources. The horizontal error
vector by itself was a little less than 2 centimeters or
0.000 02 of the altitude, or essentially 1 part in 50,000.
Block adjustment techniques were applied. The observation
equations were weighted in such a manner that the standard
deviation at the control points was essentially equal to the
standard error known to exist in the classic geodetic opera
tions as indicated by the geodetic computations.
This accuracy is very near what is being experienced in
satellite geodesy. It may be true that this accuracy closely
approaches the ultimate that can be obtained without some
major improvement in equipment or techniques, such as an
increase in image resolution by a factor of ten, together
with a corresponding improvement in the measuring engine.
The standard deviation of parallaxes was less than
3 microns. The standard error of the monocomparator was
determined through extensive tests to be smaller than one
micron. One is reminded that the least count of the comparator
is also one micron.
The small errors were achieved by virtue of the inherent
favorable characteristics of analytic photogrammetry. The
photogramme trie solution is as free as possible from mechanical
and optic defects and maladjustments of machines and instru
ments. Corrections for the distortion of the aerial camera
lens are made to the observed image coordinate values, and
no other subsequent optical distortion can enter into the
solution. If film is used in the camera, a correction for
distortion is administered for each photograph individually.
The perspective solution, being simply Euclidean geometry
and classic algebra computed to any desired number of
significant digits, retains all the inherent accuracy of the
input data without adding any errors of its own.
Operation Items
Essentially all the practical applications of computational
photogrammetry in the U.S.A. now apply the projective trans
formation equation popularized by Dr. Schmid: