block
rm was
‘Brown
“ation
berger
on -to
lens
es of
asses
by a
7e for
radial
sas on
irious
'epted
rtion
| the
1941,
lasher
n his
|, the
nally
it the
ed to
if the
1idity
mical
ld be
its
found
arity
'eren-
ed by
using
being
ed to
gular
film
ses of
1 oped
[.A.L.
cers’
953),
lard.
iction
from
asily
ished
search
n odd
tance
pt is
here.
heres
pment
ation
versy
cally
ight.
egard
'S are
(IMC)
d the
formulas with regard to only the rotational
movements and certain corrective (although partial)
approaches were developed to provide film or camera
movements during exposure by using one of several
IMC devices. The complexities has been resolved by
Ghosh (1985) through augmenting the collinearity
condition equations. This approach is fully
computational.
6.2.3.3 Related to Stereo Images
It was around 1953 that the classic analog concept
of relative orientation by way of elimination of y-
parallax evolved into the condition of coplanarity
through the efforts of Schut (1956-57) at the
National Research Council of Canada. This
condition implies that the two perspective centers
(or exposure stations), any object point and the
corresponding image points on the two conjugate
(overlapping) photographs of the stereo-pair must
all lie in a common plane. This condition is fun-
damental to relative orientation or space
intersection. Like the collinearity equations,
this condition equation is also non-linear and need
to be linearized (for computer utilization) with
iterative solutions in mind. The relative
orientation formulation developed by E.H. Thompson
(1959) showed complete elimination of trigonometric
functions with a consequent ease and speed of
computer utilization. Separately, Paul Herget in
developing a system of analytical control
extension, by using vector notation, minimized the
perpendicular distances between pairs of
corresponding rays in order to achieve a solution
for relative orientation. He employed an ingenious
trick whereby ground control equations took the
same form as relative orientation equations (Herget
and Mahoney 1957).
On the other hand, the superiority of the numerical
relative orientation (over empirical and graphical
methods) was definitively established. Also were
established the processes of improving such
relative orientations (Ghosh 1964).
Notwithstanding the analytical conditions of
collinearity and coplanarity, the on-line solutions
at analytical plotters are all developed
practically around such analogical-numerical
concepts.
The process of absolute orientation (i.e. scaling,
translating and levelling of a stereo model with
respect to a ground reference coordinate system) is
simply a problem of coordinate transformation. The
equation must be linearized before it can be used.
The method of least squares may also be used. This
approach was standard already by the early 1950s.
It was readily found that during a sequential
procedure of aerotriangulation the scale of a
previous model needs to be transferred to the next
model. This is similar to the requirements of the
analog aeropolygon method. This process was
mathematically modeled at the NRC Canada (Schut
1956-57) and is known as the scale restraint
condition. This condition implies that with regard
to a point in the triple overlap area (i.e. area of
overlap between adjacent models) the two
intersections in individual models must take place
at the same spatial location. This condition is
always used in conjunction with the coplanarity
equations.
Theoretical concepts of bi-projective
transformation (Das 1952) and the use of distances
in the object space as control for stereo-models
(Das 1973, Okamoto 1981) are purely computational
315
approaches that would prove extremely efficient in
various applications of stereophotogrammetry.
6.2.3.4 Related to Multiple Images
The application of analytical procedures on which
most discussions and efforts have been made is that
of phototriangulation. As early as the beginning
of World War II, the need was typified in the
following quote (Schermerhorn and Neumaier 1939):
"The problem of control points was and is still, to
a certain extent, the bottleneck in photogrammetric
map production". Initial efforts were with regard
to the adjustment of analog aerotriangulation.
Later efforts concentrated on fully analytical
procedures. Their classifications and historical
developments would be apparent in Fig. 6.2
A. Adjustment of Analog Aerotriangulation
Historically, the development may be noted in
terms of three stages:
Stage 1: Adjustment of individual strips along
with associated data analyses and
interpretations. The works of Thompson (1953),
Roelofs (1949) and Gotthardt (1944) give typical
indications of the initial studies. One would
notice at this stage the prolonged discourses on
the causes and propagation of random errors over
those of systematic errors. One can refer to
one single publication to typify the culmination of
this stage in the OEEPE (Organisation Européenne
d'Études Photogrammétriques Expérimentales) report
for studies up to the end of 1959 (Solaini and
Trombetti 1961). This study, initiated in 1956,
concerned international efforts at twelve research
centers and analyzed the results of some dozens of
strip triangulations by using different adjustment
procedures.
Several scientists got involved in such studies in
the OEEPE group or separately and left their marks
in numerous publications of each of them, such as,
W.K. Bachman, A.J. Brandenberger, A. Bjerhammer, A.
J. van der Weele, A. Verdin, P. A. Vermeir, J.
Zarzycki and M. Zeller. The efforts of the ISP
Commission III in this regard were very significant
(see Cassinis and Cunietti 1964). The OEEPE (1973)
publication indicates the termination of experi-
mental researches of this stage, having the
attention already passed from the treatment of
isolated strips to that for an entire block. The
highlights of this stage were: (1) Adjustment of
aerial strip triangulation was approached by using
condition equations; (2) The least squares
principle was being applied to the condition
equations; and (3) The polynomial corrections of
point coordinates were affected by considering
third order in X, and second order in Y and Z strip
coordinates. It was also felt at the end of this
stage that the measuring instruments and the opera-
tional procedures needed improvements more than the
mathematical adjustment procedures.
Stage 2: Adjustment of Blocks of Strips. By the
end of 1950s in view of the developments that
electronics brought about in the computation
processes new challenges concentrated on
simultaneous adjustment of blocks of strips, the
models of which have been formed by analogical
procedures. In this regard, apart from numerous
individual efforts in the world, the one most
significant study which would indicate the progress
is the report on the coordinated group study under
ISP Commission III on "Massif Central" polygon