233
curves
i
dR
Fig. 8 - The estimated characteristic distortion
curves and the Rollei calibration curve
Cerfificate curve
The Principal Point is estimated as intersection of
straight lines passing through the distortion vectors
making an estimation not very accurate, as can be seen
in table 2.
Fig. 9 - The estimated calibrated distortion curves
versus the Rollei calibration curve.
The test have been performed both with the non-
corrected terrain co-ordinates both with those with the
correction for relief displacement. The interpolation
accuracy was as in table 1.
Table 1-No relief displacement correction vs.
correction
Test 1
Test 2
Test 3
<7q nocorr
0.112 e-5
0.113 e-5
0.115 e-5
(TQ with corr
0.109 e-5
0.108 e-5
0.107 e-5
Hypothesis
testing
Ho
Ho
Ho
For instance in testl :
• With no relief correction Sigma-naught = 0.112
e-5 mm
• With correction Sigma-naught = 0.109 e-5 mm.
Making a hypothesis testing with a significance level
of alfa= 5%, the null hypothesis is true.
That means that the Flatness of the wall is not critical
for the present conditions (for lack of planarity
^ 1/2000 & for angle of field ^ 90°).
Table 2 Results of the estimate
DISTORTION COEFFICIENTS
Ro=20.0 mm
Ri
r 2
r
Princ. point
(mm)
ROLLEI
0.207e-4
-0.97e-8
?
0.12
-0.24
TEST 1
0.18e-4
-0.99e-8
-0.98
1.75
0.78
TEST2
0.21e-4
-0.13e-7
-0.98
0.49
-1.16
TEST 3
0.15e-4
-0.86e-8
-098
0.69
-0.63
Ro=23.0 mm
TEST 1
0.24e-4
-0.12e-7
-0.97
TEST 2
0.26e-4
-0.13e-7
-0.97
TEST 3
0.21e-4
-0.11e-7
-0.98
In this manner the found point is likely to be near to
the point of best symmetry.
The estimated distortion coefficients R1 and R2 are
highly correlated (r = -0.97-f--0.98), so that the
comparison with the Rollei coefficients is not really
meaningful.