reseau points are measured, a least-squares solution can be ob-
tained. For most TV mapping systems, the field angle is relatively
small and terms involving Pi» P5 and P3 in the above equations may
be safely omitted. -
Simulation Studies
Fictitious data was generated to test the calibration
technique under several conditions of measurement accuracy and
system constraints. Three test cases were conducted for the
RBV camera configuration (16.22° field) and three test cases
were conducted for a camera with the same focal length but a
45° field. The latter cases were aimed to study the effect of
the narrow field angle of the RBV camera in the mathematical
solution. The results are summarized in Tables 5 and 6.
In cases 1 and 2, the unknown parameters including the ,
exterior orientation parameters w, $ and K were loosely constrained.
Although the focal length could be determined to within +0.002 mm.
in both cases, large uncertaintie$ were found in the computed x
and y . Because of the narrow field angle of the RBV camera,
the wBak resection geometry presented strong correlation between
the principal point coordinates (x _, y.) and the rotational
parameters Uu, $ and'k. Case 3 showed That in order to improve
the accuracy of x and yp the rotation parameters 0, $ and K
"must also; be constrained to a high degree of .accuracy in the
solution: This problem did not exist in the 45° field camera
tested in cases 4 and 5.
Because of the narrow-field angle, lens distortions were
largely linear and could be compensated by a small error in the
equivalent focal length; as shown by the results in case 3.
Whereas cases 6 and 7 showed that for a 45? field, both the
true focal length and the actual lens distortions could be
accurately determined.