elements of exterior orientation for all exposure stations: 
Ou, X,X^, Y ,Z*; 
elements of interior orientation x, , y, , c; 
coefficients of radial and decentering distortion: K,,K5,...;P, , P... .; 
object space coordinates of measured points: X,Y,Z. 
The method differs from Aerial SMAC in that no control points are required, though 
if available they can be exercised with appropriate constraints. The theoretical 
development of the method consists essentially of the straightforward extension of 
the theory of block analytical aerotriangulation developed in Brown (1958) to 
include recovery of parameters entering the composite model of distortion developed 
in: 
Brown (1956): radial distortion; 
Brown (1965): decentering distortion; 
Brown (1971) and present paper: variation of distortion with 
object distance. 
Because the method requires no knowledge of object space coordinates or knowl- 
edge of other object space relationships, we refer to it as constituting a process of 
self-calibration. 
The process of self-calibration is best illustrated by means of a concrete 
example. Figure 8 is a reproduction of a photograph of the DBA target range 
(mentioned earlier in conjunction with Aerial SMAC). The photograph is one of 
eight taken by a DBA 480mm plate camera (18 x 18cm format) focussed at 12.5 
meters. The eight photographs were taken in pairs from four different exposure 
stations symmetrically located with respect to the target array. The pair of 
photographs from each station were inclined 45* from the vertical and consisted 
of one exposure with a nominal swing angle of zero degrees and of the other with 
a nominal swing angle of either plus or minus ninety degrees. The photographs 
were reduced by the self-calibration process in various combinations, three of 
which are indicated in Table 2 along with results from a Stellar SMAC calibration. 
Decentering coefficients from Stellar SMAC were exercised in the various reductions 
with one sigma constraints derived from the output of Stellar SMAC. All of the 
resulting corrections to decentering coefficients turned out to differ insignificantly 
from zero and so are not presented in Table 2. 
Case 1 was presented to illustrate that, as in an Aerial SMAC calibration, 
a diversity of swing angles is essential to a successful outcome from the process of 
self-calibration. Here, acceptable results are not obtained for y, and c. The 
shortcomings of Case 1 are rectified in Case 2 by the choice of a set of exposures 
exercising suitable variation in swing angle. This provides the needed improvement 
in y, and c while maintaining acceptable accuracies in x, and’ K, . In Case 3 all