Testing of Stereo Vision Quality 
By dr P. O. Fagerholm, Sweden 
As stereo-vision is of importance not only for the work in stereo- 
plotters but also for the more or less general use of stereo-photographs, 
especially aerial photographs, in planning, photointerpretation etc., the 
author has made some studies about a couple of testing methods used in 
Sweden during the last decade. 
The first method is based on the wellknown "Zeiss Prufungstafel” 
which contains a lot of circles, dots and small silhouettes arranged in 
three dimensions. The testpicture is to be viewed through an ordinary 
pocket stereoscope. Thirteen "problems" have been selected — the per- 
son to be tested has to group in each problem 2 to 5 details in depth or- 
der. A correct answer to a problem gives certain points, which have 
been based on the results of the first 300—400 tests. The maximum- 
sum is 10 points. 
'The second method is the American "Wirt-test" which uses the pola- 
roid-vectograph-system to separate the two images which are superim- 
posed on each other. Maximum-sum is 14 points. 
These two test methods are of quite different nature and contain, 
each of them good and less-good features. The sum of the two tests has 
been regarded as giving a more true result, consequently. 
The author has analyzed about 1.100 of the estimated 3—4.000 tests 
performed in Sweden during the last decade. The persons tested have 
been students at the Royal Institute of Technology and the forest 
schools as well as participants in various courses for surveyors, carto- 
graphers, dentists etc. The distribution; however, in various ages, types 
of work etc. is rather un-equal. The testing procedure has also been 
varying to a certain extent. Consequently, the author has not been in- 
clined to proceed to far with the analysis. 
The following approximate results may be of interest. Each result 
should be accompanied by several comments but space does not allow 
that: 
1. The distribution in the "Stereoscope-test", the "Polaroid-test" and 
the "Sum-test" is shown in Fig. 1. A jump can be found at 10 points in 
the last mentioned test and similar jumps in the two basic tests. The 
separation between the good results is obviously far to small, 
9 
“wl