Ability to Maintain Correlation
The automatic stereoplotting device keeps 'locked in" for the quite
rapid speed of the plotting table. During the experiments the plotting table was
moved from point to point at a speed of about 10 cm per second for several hours,
but the device never lost contact with the required correlation.
Accuracy of Relative Orientation
It was intended to test the accuracy of the relative orientations on the
automatic stereoplotting device by observing the y-parallaxes visually at the five
orientation points. However, it turned out that even experienced operators could
not ascertain any residual parallax at these points. Therefore it was decided to
determine the accuracy of relative orientations indirectly from the differences in
elevation readings between consecutive orientations.
Two different kinds of tests were made. First, one orientation element
of the right projector was changed one by one, and secondly, all the elements were
disturbed at the same time. In both cases the elements of the left projector were
kept unchanged except that the device also used Xj when it was eliminating the
y-parallax at position 6. In each case the procedure was
(1) read the elevations of the seven points shown in Fig. 6,
(2) disturb the orientation element or elements,
(3) allow the automatic stereoplotting de-
vice to recover the orientation, and
(4) re-read the elevations.
eo
6 From the discrepancies between
the initial readings and the readings after
the recovery of the orientation, the corres-
ponding differences in the values of the
orientation elements were calculated by the
method of least squares. Since the influence
of errors in X| and X5 are identical these
two could not be separated and so the ele-
ments X, £1. $2, and w, were used in the
+ | ef +2 final analysis. The results are shown in
Table I.
As weighted averages the follow-
ing deviations were obtained:
c
Ax, = 10.9
c
Ag, = 41.4 C
average = *t0.8
°3 Ad, = +0. 1 | s
zd 27,10.
c
Aw, = 1.7
