252 
Envelope data may seem to come from nowhere. Actually, it is data present in 
the images that has not been used by previous correlators. Envelope data is 
particularly useful under certain conditions. For example repeating patterns, 
such as a plowed field or an orchard, tend to confuse stereo correlators even 
more than they confuse the eye. A false match between one furrow and its 
neighbor can introduce height errors of multiples of the furrow spacing times 
B/H. With envelope correlation low frequency data in the form of contrast or' 
spacing differences provide a lower band error that can override the false 
match error regardless of the clarity and contrast of the furrows or trees. 
Another application for envelope correlation, perhaps without any direct sig- 
nal correlation, could be the matching of side looking radar imagery taken 
from different angles. Such material is not normally correlatable owing to the 
incoherence of the scatter patterns. One would expect the scatter pattern en- 
velopes to be coherent however and the correlative matching of side looking 
radar imagery should be possible. 
3 CONCLUSION 
The automation of photogrammetric plotting has developed at a slow and some- 
times unsteady pace for many years. Stereo correlation has presented a series 
of technical problems that have required unique solutions having virtually no 
application except in topographic mapping. It has been difficult to justify 
the investment required to develop solutions in view of the small number of 
instruments likely to be sold. In commercial terms the two most successful au- 
tomatic plotters to date have been the B-8 Stereomat and the Gestalt Photo 
Mapper and only about a dozen of each have been built. 
There has been little activity in other industries contributing directly to 
the cause of photogrammetric automation except for the tremendous advance in 
small computer technology. The situation is rapidly changing as robotization 
sweeps through the world's manufacturing industries. The needs of robotic vi- 
sion creates a huge demand for new systems and methods, some of which will be 
usable directly in photogrammetric applications. For the first time, a wide 
variety of sophisticated yet inexpensive tools are becoming available to the 
designers of automatic stereo plotters and interpreters. We believe that the 
swelling wave of robotization will affect the mapping industry as profoundly 
as it is now affecting the automobile industry. 
Technology now exists that can extend the performance of stereo image 
correlators for photogrammetric plotting to the physical limits of the input 
material and at a very high productivity. Very large scale integration chips 
and the newer semi-custom devices greatly reduce the design time and equipment 
cost compared with earlier methods. The result will be a cost effectiveness in 
the next generation of automated photogrammetric instrumentation enormously 
higher than we have come to expect from the current generation.