elements cannot
ystematie para]
The relative Orientation
any required degree of
"ientation of the Camera
is not the Case: In the
°curacy is reacheq, This
ble to reduce the Number
cost of photogrammetyie
be considerably cheaper
' is simpler, It therefore
neil of Canada a method
» the Ferranti electronie
ow be described in some
ding points on a stereo
oint as the origin of à
ral types of errors,
À grid plate is placed
me of exposure. The c-
the nearest grid inter-
d co-ordinates of thes
S. The radial distortion
correction for radial
1 can be measured and
rtion if the photographs
umptions may be found
tria IX 1952—53 Nr. 3.
is eliminated by radial
vertieal photographs.
h curvature:
e strip to a plane. For
'h as one tenth of à
btained, the map co-
n is most conveniently
n. An arbitrary initial
€ mage
aXes are
The second type of condition concerns ground control points.
Jusly exceed the capacity of a medium sized electronic computer
lie first photograph of a strip and to compute the absolute orientati
jq. Generally more points will be measured in eac
\inecessary to establish the orientation. This will be
Jutly to increase the accuracy.
Jmarately.
{imltaneously, using all available heights.
Tri
{al trangulation has been developed and applied to instrumental triangulation. This
3
im mày be chosen. In this system the absolute orientation of each photograph is
gs
ge Bp orientation is expressed by six elements, e.g. the three rectangular co-
mates of its projection centre and three rotations that give the position of the photo-
ym o with respect to the axes of the co-ordinate System.
p six elements must be determined from the conditions which
ssonding image points must fulfill There are two types
onem ding image points in two photographs must i
m corresponding image points in three photographs
ga the rays through
of conditions. Firstly rays
ntersect. This implies that
must intersect in one point.
The given co-ordinates of a
mind control point must satisfy the equations of the rays through the corresponding
image points.
Conditions of the first type express relations. between orientat
sve photographs. Therefore, they determine primaril
slows that the absolute orientation of e
ndependently.
To determine the absolute orientation either one of two me
The theoretically best method is to compute the elements of absolute orientation of all
doographs directly and simultaneously. If a redundant number of co-ordinate readings
: available a rigorous method of adjustment such as the method of least squares can be
gplied. This method involves the simultaneous solution of as many equations as there
me elements of absolute orientation, ie. six times the number of photographs. Even in
mil mapping projects this will amount to a few hundred equations or more. This may
or else require an
is method does not seem practical yet.
lowed is to assume an arbitrary absolute orientation for
on of each following
e relative orientation
ion elements of suc-
y their relative orientation. It
ach separate photograph can not be established
thods may be followed.
winate amount of computing time. Therefore th
The alternative that must be fol
ibtograph in succession. This absolute orientation is established by th
tech photograph to the preceding one and scaling of the resulting model. The relative
mentation may be established by making five pairs of corresponding rays intersect. The
muting model is then scaled to the preceding one by making one height in the two models
h overlap than the minimum that
done partly as a check on errors,
The relative orientation and the Scale are then adjusted
The six elements of absolute orientation can also be computed simultaneously. Four
Tuis of corresponding rays may then be made to intersect each other, two of which at
tights determined in the preceding model. If more than four p
airs of corresponding
mage points have been measured an adjustment is carried out
for all six elements
This method has its disadvantages. Errors
readings result in errors in the relative orientation and thus in
kormation of the model. If two or more heights established in the preceding model are
il, the relative orientation will also be affected by errors in this model. This will
rease the accuracy. Also the triangulation result will depend on the model used to
I the co-ordinate
J'ut the triangulation.
The method of independent relative orientation and scaling should therefore be used.
After orientation of the photographs, co-ordinates of the measured points are computed
nersecting rays from corresponding image points. Finally these strip co-ordinates are
lansformed to the required terrestrial or map co-ordinate system.
As yet no possibility of adjusting the strip-triangulation as a whole has been provided.
I the Photogrammetric section of the National Research Council an airborne method of