5. DISCUSSION AND CONCLUSION 
A critical study and comparative analysis of past and current 
methods of analytical aerotriangulation illustrates clearly the merits of 
additional research oriented toward improving present methods or introducing, 
a new approach. The new treatment presented in this paper has the general 
objective of developing new concepts suitable for use in aerotriangulation. 
The single photograph is promoted as a basic element and equal 
overlap in all directions is advocated. It is likely that objection may be 
raised against the possible increase in the cost of photography. With 60% 
side-lap, about twice the number of photographs would be required. However, 
even if the cost of photography is assumed to be double the normal amount, 
it still remains as a small fraction of the total cost of the mapping opera 
tion. In addition, this small increase in cost may well be counterbalanced 
by the expected reduction in required ground control and increase in accu 
racy. Furthermore, the additional number of photographs is only used in the 
triangulation phase, while only photographs from alternate flight lines are 
needed for compilation. 
Square sub-blocks are used in place of the conventional photo 
graphic strips, and two methods of sub-block triangulation are presented. 
The first method, sub-block extension, is a direct development of the 
present method of strip cantilever extension. It has the advantage of ex 
hibiting no model discontinuities, less propagation of errors and less 
pronounced model deformations, as compared to cantilever extension in a 
strip. This is due to the fact that sub-block extension is essentially in 
all directions, while strip cantilever extension is only in one direction. 
The second method, sub-block best-fit, bears some resemblance to 
the present technique of bridging. The idea of sub-block relative orienta 
tion is introduced as a more general concept, of which relative orientation 
of a stereo-pair is but a special case. A best-agreement among the sub 
blocks and a best-fit on the control is achieved by applying a method of 
successive relaxation. The method of sub-block best-fit, therefore, 
eliminates error build-up and allows for harmonious distribution of resi 
duals between sub-blocks containing control. 
25