C» 
e 
-— 
  
  
+ << 
— 
| 
SUNOHNBONNNU LOU 
olo 
on DD 
  
| No ro m 
on 
For 
radial distortion resulting from the process of self-calibration should have 
ideally been insignificantly different from zero. Instead, these coef- 
ficients alone yielded corrections amounting to a-difficult-to-accept 
value of 30um at a radial distance of 140mm. Moreover, it turned out 
that these coefficients could not be suppressed in repetitions of the 
bundle adjustment without severely degrading the results. Therefore, the 
effects had to be regarded as having a real physical basis. To explain 
the phenomenon the following hypothesis emerged. The results could be 
attributed to an actual physical change in the lens itself due to the 
extreme coldness of the air outside the airplane at the average flying 
altitude of 6000 meters above mean sea level (over Vermont in mid May). 
This resulted in a pronounced temperature gradient between the front and 
rear elements of the lens, thus altering the distortion of the lens to the 
value recovered by the bundle adjustment with self-calibration. The large 
diameters of the front and rear elements of the super wide angle lens 
tended to support the plausibility of the argument. At the time, the 
writer was quite satisfied with the thermal gradient hypothesis. However, 
it was incorrect, as the consequences of the project in Atlanta were 
subsequently to establish. 
THE ATLANTA PROJECT 
In March 1975 a small block of 28 photos in 4 strips was flown 
over the City of Atlanta in a pilot project to test the practicability of 
photogrammetric densification of geodetic nets in an urban environment. 
Atlanta is heavily wooded in residential areas and is quite hilly, thus. 
making it a rather dubious candidate for photography taken by a super wide 
angle camera. Nonetheless, the same Zeiss RMKA 8.5/23 camera as was used 
in the Vermont Project was also used in the Atlanta Project, one specific 
objective of the test being to determine the operational suitability of 
the super wide angle camera in urban applications to densification. The 
choice of the super wide angle mapping camera (f=85mm), rather than the 
more conventional wide angle mapping camera (f=150mm), was governed by 
the following well-known considerations. When both cameras are flown at 
the same scale, they yield essentially equivalent accuracies in planimetry. 
However, because of its more favorable base/height ratio, the super wide 
angle camera can theoretically be expected to produce heights that are 
nearly twice as accurate as those produced by a wide angle mapping camera. 
This means that once the accuracy for height has been specified, a wide 
angle camera must be flown at twice the scale as a super wide angle mapping 
camera in order to produce equivalent accuracies in height; this in turn 
means that the photographic block generated by the former will contain four 
times as many photos as the block generated by the latter. Accordingly, in 
theory at least, the super wide angle camera enjoys an enormous potential 
economic advantage over the normal wide angle camera. The problem has been 
that the theoretical advantage of the super wide angle camera seems rarely, 
if ever, to have been realized in practice. Investigations reported in 
Ackermann (1974a) indicate vertical accuracies of the super wide angle camera 
actually to be inferior to those of the wide angle camera for photography 
.Of the same scale. Presumably, this shortfall is attributable to systematic 
errors that, hopefully, could be removed by the process of self-calibration. 
Another objective of the Atlanta Project, then, was to establish whether or 
not this is the case. 
The Atlanta Project was a joint effort of DBA Systems and 
Airborne Data Inc. in conjunction with the City of Atlanta and the Georgia 
Department of Transportation (DOT) all operating under the auspices of the 
-19-