Appendix I. Computation of the closing errors.
As already observed in chapter 4 several methods of computing the closing
errors of a strip are conceivable; methods which are for the most part instrumental
or principally numerical. This Appendix merely gives an example in which it is
assumed that the triangulation is carried out in a stereoscopic instrument.
a. Observations.
After the relative orientation of the first pair of photographs — 1,0 has been
finished the model obtained is set to ground control. For that purpose the lateral
and longitudinal tilts are adjusted and bz , has generally to be set to a value
differing from zero, bzo being kept zero:
b2— 390 bzo — 0 (61)
The height scale is set to give “true” heights.
When the complete machine coordinates of the ground control points have
been measured, the third photograph is inserted in place of the first one. The
relative orientation is made and the scale is transferred as indicated in chapter 1,
after which the elements of orientation are read and the complete machine
coordinates of some minor control points are measured. In a similar way all
photographs of the strip are joined successively always keeping bz; = 0.
Arriving at the last model n, n-- 1 we read its elements of orientation and
the abscissa x, of the x-transfer point after which the model is set to ground
control, keeping bz, = 0 and leaving the height scale unchanged. This is done
by adjusting the lateral and longitudinal tilts, the base length (for scale) and
bzn+1. Some of these new elements of orientation are read:
On, Qn; bn +1 and bz n+1 zÉ 0 (62)
After the height scale is adjusted to read true heights, the complete coordinates
x y z of the ground control points and the abscissa x, of the x-transfer point
are read.
The above procedure shows two exceptions to the general rule that bz should
be zero: the absolute orientation of the first model and of the last one.
b. Computation of w,, wg, we and w..
Since the first model is set to ground control its lateral tilt does not need
correction:
A mo 0
Evidently the correction Aw, to the lateral tilt w, of the last model should
be such that:
Wn + A wy, = On
whence according to (10):
Woo = À On — À DO = Wn — On (63)
For similar reasons we have:
dbo = 0 and bn = 1 + db n+1 b, +1
whence, referring to (14):
we = dbn+1—dbo = baba cent (64)
18