Two translations dx 0 and dz 0 of the image coordinates.
One scale change dc, referred to the principal distance.
One rotation dx around the camera axis.
Two rotations dcp and dco around orthogonal axes through the (exterior)
perspective center.
The differential formulas relating these regular errors and the corresponding
errors of the image coordinates are well known, see for instance Haller1 1960 a.
(1)
The discrepancies between measured and given image coordinates are defined
as follows.
dx x meas X«:.
(3)
(4)
meas
given
dz = z.
meas
given
The expressions (1) and (2) are written as errors of the image coordinates,
caused by differential regular errors of the measurements in the comparator
(dx 0 , dz 0 , dx), of the principal distance c of the camera, and of the two orienta
tion angles (p and co of the photograph. Though either errors or correc
tions may be used in the basic differential formulas, we will here determine
the corrections of the six preliminary data. Therefore the signs of the right
side of the expressions (1) and (2) are changed. Furthermore, since redundant
measurements are assumed to be available, residual corrections v x and v z are
assumed to be added to the measured coordinates x and z. Hence the residual
correction equations are obtained in the usual manner:
(6)
Note that dx and dz still are defined according to (3) and (4).
For the adjustment of the discrepancies, normal equations are formed in
the usual way and solved. The corrections of the preliminary values of the
six parameters, the standard error of unit weight and the weight- and correla
tion numbers for the determination of the error propagation from the standard
error of unit weight are obtained from the solution. Finally, corrections can
be computed to the measured image coordinates in arbitrary points according