3.4 Determination of the Correlation of y-parallaxes
A y-parallax is defined as :
p y =yi ~ y2
According to the general law of error propagation it follows that:
Qpy p y — Qy x yi + Qy 2 y 2 — 2 Qy x y 2
Therefore the squares of the standard errors are:
s ° 2 QpyPy — s ° 2 (Qy i y 1 + Qy 2 y 2 2 Qy x y 2 )
Qyiy 2 =
'y x yi
Qy 2 y 2 — S f
Assuming that Q y iY ^ — Q y >y = 1 the correlation number is :
-y l y 2
: 2s 2 0
According to sections 1.62 and 3.3 the following mean values are found:
s n =7 microns
5 =5 microns
Therefore Q yi y 9 = 0.7
This is an expression of strong correlation between the y-coordinates of overlapping photographs.
Qy y
The correlation coefficient is defined as r y _r^ 2 =
v / Qy 1 y 1 Qy 2 y 2
and is in this case, under the preceding assumptions, identical with Qy^ = 1 — s Py 2 : 2s 0 2
4. THE ABSOLUTE ORIENTATION AND THE COORDINATE DETERMINATION
After the relative orientation in a stereoscopic projection plotter the model coordinates are determined
by setting the measuring mark and reading the coordinate scales. The precision of this setting operation
should be known and can be determined from replicated or repeated settings and coordinate readings
only. The accuracy of the model coordinates depends upon the precision and also upon the accuracy of
the preceding operations especially upon the errors of the relative orientation, which cause model defor
mations.
K. Torlegdrd, MSE, has conducted a series of tests of the precision of the settings in model points
under various circumstances. These tests have been made in cooperation with medical experts and
have included tests of the effect of alcohol and smoking on stereoscopic setting of measuring marks.
From K. Torlegard’s publication some of the tests results are summarized here. See 4.11 and 4.12.
It is evident that the precision of the settings usually is so high that the contribution of the setting
errors to the final errors of the photogrammetric procedure is generally of limited importance. Moreover,
the precision can be increased by replications and repetitions of the settings and determination of the
averages, as the standard deviation of the average decreases with the square root of the number of repli
cated or repeated settings. In empirical relative orientation, drawing of planimetry and contouring,
however, only one setting is really being used.
Replicated measurements often prove to be partially dependent on each other and, therefore, repeti
tions can be expected to give a better average.
4.1 The Precision of Model Coordinate Measurement under Special Conditions
4.11 The Effect of Alcohol C onsumption
The precision has been determined as a standard deviation of one setting from the results of 25
replicated settings in photogrammetric models with 15 series measured during a day. Alcohol was
administered to two operators while two others with no alcohol intake served as a control group. Read
ings and recordings were made by assistants. On another day the operators reversed roles. The effect of
two different types of alcohol intake was investigated, viz.
