197
SUB-COMMISSION IV-4
FUNDAMENTAL QUESTIONS IN RELATION TO
CONTROLLED EXPERIMENTS
President: Prof B. Hallert Associates: Mr L. O. Hosen; Mr C. O. Ternryd
Presentation of the General Report of Sub-Commission IV-4
in the Meeting held on Friday, 9th September, 1960
by the President, Prof B. Hallert
In a few words I shall introduce the work
and results of sub-Commission IV-4 during the
period 1956—1960. Since all details can be ob-
tained from the printed report I will make a
very brief summary here, and mainly to avoid
misunderstandings will especially point out
some limitations of the methods and procedures
which have been applied in our sub-Commis-
sion.
We have concentrated upon the investigations
of the fundamental operations of photogram-
metry, and I have tried to determine some im-
portant regular errors and to estimate the
magnitude of the residual irregular errors of the
central projection of photographs. This is en-
tirely in accordance with the usual procedure
in, for instance, geodesy. In particular, the in-
struments and models have been tested by our
Commission since we have not heard of a possi-
bility for testing the photographs themselves.
We have investigated the propagation of the
irregular errors through the entire photogram-
metric procedure, and have estimated errors of
the final results expressed as mean values of the
standard errors of the co-ordinates in planimetry
and elevation. This data we have estimated
could then be directly checked by a comparison
with the mean square values of the discrepancy
between the photogrammetric and the geodetic
co-ordinates according to sub-Commissions
IV-1 to IV-3.
Since no exact agreement can be expected in
such a case, the significance of the differences
between theoretical and practical values have
been judged with the aid of statistical tests, in
particular confidence limits according to well-
known principles and according to a certain
confidence level which is chosen as 595. In all
cases, an astonishingly good agreement has been
found between theoretical and practical results,
and it seems important that some theoretical
relations of significance for photogrammetric
practice have been applied.
It seems possible to construct certain speci-
fications for photogrammetric work concerning
Archives 5
the fundamental operations, in particular from
the x and y parallaxes in oriented models, but
it must be emphasised that the fundamental
operations (1) and (2) — the plotting of the
photographs in the plotting instruments, in other
Words — have to be especially checked first.
There have been some very regrettable mis-
understandings in this respect since the Stock-
holm Congress when these principles were
published, in spite of the clear reservations
which were made there. I will here repeat my
own words in the paper on the grid method and
Y-parallax method for the determination of
systematic disturbances in aerial and terrestrial
photographs, which was published at the Stock-
holm Congress in 1956. I quote again:
"Finally, it has to be emphasised that the
grid method gives more accurate and correct
information on systematic disturbances than
the y-parallax method. The latter method is
particularly limited by the considerable cor-
relation between the two pictures. There may
be distortion effects in the two pictures which
cannot be distinguished in this way. Neither
of the two methods can give the absolute
determination of the orientation under as-
sumed conditions."
The expression "correlation" may perhaps
need some extra work. Since a parallax is al-
ways defined as a difference between two image
co-ordinates which may be enlarged, it is quite
evident that any source of error which has a
similar effect or an equal effect upon the two
co-ordinates will not show up as a parallax. This
means, for instance, that the influence of the
earth's curvature, symmetrical refraction, some
kind of film supporting glass plates in the
camera — as in the English cameras, for instance
— certain curvature of the locating bag or nega-
tive glass plates, certain shrinkage of the nega-
tive material, in particular a fine shrinkage,
cannot be determined from parallax methods.
The image co-ordinates in such a case become
correlated and they also become correlated via
the element of their relative orientation, in