Lcal,
idies
arial
ough
Fool.
(d to
Lt so
[thin
) the
1racy
1ates
[t ig
opic
>the
sural
r its
vhich
with
ope,
^ de-
ated
age.
d be
ver,
into
- 243 -
In practical air photo interpretation there are cases where
model deformation will not be found disturbing at all: When
measuring short distances, minor level differences, small areas
or volumes, distortion of the stereo model will hardly have any
noticable effect on the measurement result.
On the other hand, there are situations where model distortions
will introduce considerable errors, for instance when measuring
elevation, route profiles or slope angles. In order to make
allowance for these differences in measurement problems there
are different coordinate transformation programs for the
STEREOCORD, designated COORD A, COORD C, and COORD D (Fig.4).
In COORD A $=wWw=0 is assumed; in other words, no allowance is
made for photograph tilt, just as in measurement with a parallax
bar. The advantage of this approach is that the user may start
measuring in the stereo model right away without recourse to
the orientation programs, no control points being required.
However, accuracy will only be satisfactory when measuring
short distances or small areas or volumes. As soon as longer
distances are involved, model deformation will introduce con-
siderable errors. As an example, Fig.4 shows a test model flown
at 2,600 m, with photograph tilt around 18 - 28, The diagrams
show the distribution of terrain-coordinate errors over the
model surface. For greater clarity, the arrows are exaggerated
by the factor 8 in relation to the model surface.
COORD A COORD C COORD D
and parallax bar Sing=¢; sinw=w sing ;sinw
P= wW=0
cos@=cosw=1 COS Q ; cos w
50m
deviations exaggerated by factor 8
50m ^X
Fig.4: Accuracy of coordinate transformation using the
transformation programs COORD A, COORD C, COORD D with
a test model (camera focal length 153 mm, flying height
2,600 m, Q1 -1.38, 92 = +1,08, (1s 1,48, 02 = 1.72).
For more accurate computation of terrain coordinates, the
COORD C transformation program has in the past been available
for the 9810 A calculator by Hewlett Packard. Since standard
equipment of that calculator did not include trigonometric
functions, allowance for photograph tilt was made to a first
order approximation in COORD C. As a result, the accuracy in
the test model was increased by about the factor 5 compared to
COORD A.