122 CONVERGENT VERSUS VERTICAL PHOTOGRAPHY, DISCUSSION
instruments. How many cameras lying around
the world are really calibrated regularly? It is
one of the advantages of the big services like, for
instance, the Geological Survey and the French
Institut Géographique National, and there may
be others, that they get such excellent results not
because of the marvellous quality of their equip-
ment but because of the careful considering of
each link in the total chain of the procedure
from the moment that the plane takes off until
the moment that the photograph arrives in the
restitution instrument. I have the impression
that so far that has been overlooked by many
services.
I have said before the war many times that if
people ask me which instrument was the best in
your institute, the Autograph A5 or the Stereo-
planigraph CS, I said at that time that both are
good but much more important is the question of
the quality of both ends, that is the photographs
you put in it and the man who is behind it. That
is still the point.
We have also heard something here from Mr
Blachut about personal errors which go above
and beyond the precision of the equipment, and
We are in a position at present where we have to
talk about those sources of errors.
Mr T. J. BLACHUT: I would like to ask Mr
Thompson one question and also maybe give a
very short comment. I will start with the com-
ment. It is, in our opinion, not only the question
of calibration of cameras which must be, of
course, very carefully done within certain
intervals, but also the question of processing
the film and conditioning the film. These re-
quirements are at least equal in importance as
the calibration of the camera. We proved that
if, for instance, photographs are made imme-
diately at the beginning of the strip of film and
so on, one cannot expect reliability and homo-
geneity of results. That also must be observed.
As to the question for Mr Thompson, 1
would like to ask him if he refers to cameras of
identical performance when stating that the
results are 20 per cent better? Because that is,
of course, very important.
Mr M. M. THoMPsoN. I would like to
emphasise what I mean by 20 per cent better.
In 30,000 square miles of mapping you do not
get an answer, the accuracy is so much, you get
a probability curve, and we got our probability
curve in the form of a cumulative frequency
curve, the S-curve. In plotting these probability
curves the curve for a convergent, that is a com-
posite curve for convergent photography, is
displaced about 20 per cent. In general when I
say 20 per cent it is the general amount that one
probability curve is better than the other.
Your question has to do with the use of iden-
tical cameras, in taking the twin photography.
The cameras, of course, are not identical, the
tests are not laboratory tests, our tests are
operational tests in which we use cameras that
have been processed and approved in the
calibration laboratory under Mr Bean, who
may wish to say something additional about
this. Every camera used in this work has gone
through our optical laboratory and been ap-
proved, but the cameras are not identical, of
course, so we are not comparing identical things.
But we are pragmatic about it, we are looking
for results which we will be getting in operation,
and of course these are not theoretical results
but practical results. Mr Bean may wish to add
something on that point.
Dr SCHERMERHORN: First I want to ask
members of the panel if any one of them can
answer the question from Professor Hallert.
I would like to say a few words about this
problem; it is one of those facts we meet in
photogrammetry every day, viz that there is
very little known really about the fundamental
points and problems in photogrammetry. We
are all measuring, but many of the problems
raised at present by Professor Hallert still re-
quire a clear-cut answer, based on sufficient
statistical material. But to get this is such a
tremendous job, which gives so little pleasure
to the operators that hardly any of us has started
it. It is one of the aims of Commission F of the
OEEPE to work together on this problem. This
question raised by Professor Hallert seems to
be one of these unsolved problems because also
the members of the panel have no answer, what
[ expected. Professor Hallert himself is the man
who can give the answer.
Professor B. HALLERT: As a matter of fact
we have done some experiments from a high
tower, we have taken photographs with oblique
cameras. We have made a perspective grid on
the ground in order to have the points on which
one circles around the principal point because
then the normal equations, you know, for the
determination of the inner orientation can be
made in advance. In that way we can distinguish
between some important systematic errors and
obtain the irregular errors as a standard error
of unit weight of the image co-ordinates. I can
tell you that the first experiments have given
the same magnitude for oblique cameras as for
verti
whi
com
moc
floo
to
Thc
rap!
20
pho
in |
met
ord
gen
mo
mo
ph
eve
are
dat
cor
acc
anc
1/1
are
ver
ac
tin
aei
50
the
vo
Tr
nà