142 
Discussion 
image?» Now the question is put in this way: »What is the contrast transmission 
factor (in per cent) of an image-forming system for a sinusoidally varying 
pattern of a certain number of lines per millimeter?» These ct-factors describe 
the whole ct-function in terms of the number of lines per millimeter as abscissa. 
The ct-functions have a very simple connection to object contrast, and it is 
shown how certain image disturbances, as unsharpness due to motion, are calcu 
lated into the same function. 
A short review and bibliography is given of recent designs of practical appa 
ratuses for obtaining these ct-functions, made in the United States in Great 
Britain and in Sweden. Some of these equipments obtain a function which also 
includes the photographic emulsion. 
Interesting questions arise as to the relationships between the shape of the 
ct-function and the practical information required from a picture, e. g., by an 
image interpreter. Some results of a pilot research are given. 
E. lngelstam tfc P. Hjelmström: Neue Richtlinien zur Angabe von Abbildungseigenschaften 
mit Verwendung von KT-Funktionen, und ihre Beziehung zu praktischen Fragen der 
Photographie. 
Die Verwendung von Kontrasttransmissionsfunktionen (KT-Funktionen) wird wahrscheinlich die 
Angaben von Auflösungsvermögen ersetzen. Man fragt hier nach der Prozentzahl, womit das abbil 
dende System den Kontrast sinusförmig varierender Strichgitter einer bestimmten Anzahl von 
Linien pro Millimeter wiedergibt. Die KT-Funktionen geben diese Prozentzahlen als Funktion der 
Anzahl Striche pro Millimeter an. Die Beziehungen zum Objektkontrast und zu gewissen Störungen 
der Abbildung sind mathematisch einfach berücksichtigt. 
Eine kurze Übersicht von Apparatenkonstruktionen für KT-Aufnahmen wird gegeben. Wichtige 
Fragen entstehen betreffs der Beziehung zwischen der Form der KT-Funktion und gewünschten 
praktischen Informationen aus den Bildern; einige vorläufige Resultate werden gegeben. 
Le Président ouvre la discussion sur cette conférence. 
Mr. P. Hjelmström: Here on the screen is a graph contrast rendition at differ 
ent frequences and of information obtained from image interpreters. I want to 
point out the likeness between this last curve and the low-frequency-curves, 
which correspond to the basic frequency of the target as imaged. There are 
difficulties which have to be overcome since the lens acts as a low-pass filter. 
So the images do not always give all the information which could physically 
or optically be given. So we have to change a little this procedure and I hope 
that in the future we will be able to refine the mathematic solution of corre 
lating the ct-function with the information obtained by the interpreters. 
Mr. P. Hjelmström projette un diagramme du rendu de contraste pour différentes fréquences et 
une courbe d’information obtenue par l’interprétation de l’image. Cette dernière montre une ana 
logie avec les courbes à basse fréquence. Les difficultés à surmonter viennent de ce que l’objectif 
agit comme un filtre passe-bas. 
Mr. R. W. Fish: I only want to ask Mr. Hjelmström how were the values for 
information content in the photographs obtained? 
Mr. P. Hjelmström: The amount of information was calculated by measuring 
the probability of the incidence if the interpreter got the right answer to his 
problem. He had to identify some objects and, since there were several inter 
preters, one could calculate the probability of the correct answer; taking the a 
priori information into account is done by estimating the probability that the 
interpreter should have such and such object lying before him and that could 
not be done otherwise here than by estimating the same figure for the whole 
trial. So it came into the operations as a constant. Actually the ordinates of 
that lower part of the slide are a function of the probability that the interpreter 
should tell the right answer. 
Mr. R. W. Fish demande comment ont été obtenues les valeurs qui caractérisent l’information 
contenue dans les photographies. Mr. Hjelmström indique qu’on a utilisé plusieurs interprétateurs 
dont l’un calculait la probabilité de réponse correcte dans l’identification de quelques objets. Les 
ordonnées de la partie inférieure de la courbe sont fonction de la probabilité qu’a l’interprétateur 
de donner une réponse correcte.