(385) 
the horizontal line appears on the panoramic views. The y and z directions of 
the plotting instrument are interchanged. 
The orientation is performed in the plane having the best profile, generally 
midway between the projectors where the photographs have the greatest com- 
mon overlap. The ß (x-tilt) correction is assured in this plane. 
The parallax à is removed by bz at two points in the plane z/y —t and in 
another plane z/y=t" as far apart as possible from each other. The linear laws 
of variation of parallaxes relative to x and t easily furnish the values by: and 
bys for 4 at the points x —b, z—0, and at xo, z—o. These quantities are intro- 
duced successively as bz components and the parallaxes are removed in the 
planes x=b and x=o respectively by correcting a1, and a». After returning the 
bz to zero, the remaining parallax is removed in the same planes by correcting 
y, and y», which corrections are later repaired for more distant points if it is 
necessary. 
DETERMINATION AND CORRECTION OF THE EFFECTS OF LOCAL 
TRANSVERSE DEFORMATIONS IN PERSPECTIVE PROJECTIONS 
by 
George Poivilliers. 
Abstract. In the orientation of aerial photographs in stereo triangulation 
one sometimes encounters deformations which seem to be attributable only to 
anomalies of atmospheric refraction. The transverse deviations produced by 
these anomalies are diagnosed by a method that permits their correction. 
The theory of relative orientation is based on the exact angular similarity 
of the projectors and photographs to the bundles of lines joining the perspective 
centers and the various objects on the ground. 
However, errors in similarity are apt to exist. They arise from, for example: 
atmospheric refraction, a faulty correction for lens distortion, dissimilarity in 
the lenses of the projectors, temperature differences between the aerial camera 
and the mapping which behaves like varying factors of the principal distance 
and image dimensions. Those faults which have the same pattern in all the photo- 
graphs of the same flight introduce systematic errors. These errors are in general 
very slight and are not detected in an isolated pair, if the aerial photography 
and plotting are conducted carefully. On the contrary, they are involved in the 
“chain” of errors of position of the projectors that accumulate from one pair to 
another and become rapidly important; but the systematic character of these 
errors permits one to establish rules for their correction. 
However, sometimes one notices appreciable errors even in an isolated pair, 
where noted accurate control shows that the errors cannot be due to instrumental 
faults, nor to a halo, nor to movements of the vegetation between the two 
camera stations. Those errors, known to be present in film cameras seem to be 
equally apparent on glass plates where film distorion doesn’t exist. Further, in 
stereo triangulation, one discovers from time to time large breaks in the distri- 
bution of the errors that depart abruptly from the rule of systematic correction. 
It seems to us that these local deformations and breaks in aerial photog- 
i 
y 
s 
® 
* 
i 
i 
; 
: