6 
Gleaning out trends from the impacts of such a bewildering torrent 
of new technology is not easy. Yet, there are clear trends. There is, 
for example, a clear trend toward reduction of interconnections. Inter 
connections are a notorious source of trouble, not only physically, but 
also conceptually in terms of logic and timing. Similarly, and perhaps 
not totally independently, there is a trend toward moving functions from 
hardware to firmware, and from firmware to software, whenever the speed 
requirements can be met. (An exception to this is the computer main 
frame application, where decreasing hardware cost has led to a trend 
of adding hardware to reduce software development costs.) Steadily de 
creasing memory module cost supports continuation of these trends. 
A direct result of the availability of microprocessors and their 
popularity as means to implement random logic has led to another clear 
trend: use of distributed processing. In its purest sense, distributed 
processing means incorporation of computers into data originating stations 
"Smart" terminals, point-of-sale terminals, and "intelligent" instruments 
in a data collection system are typical examples. Similar concepts are 
favored for automation of manufacturing. The kernel idea is to perform 
whatever processing is possible - and reasonable - near the point where 
the data originates. The results are then transmitted to the next higher 
processing level, sometimes over significant physical distance. However, 
long distance transmission is not an element in the basic concept of 
distributed processing. Thus, the principle of distributed processing 
could apply in photogrammetry to individual pieces of equipment, or to 
the entire photogrammetric data processing system. 
Multiprocessor systems, organized in nets, fans, rings, or other 
configurations and operating in parallel or serial fashion distribute 
processing resources by task rather than location. In a wider sense, * 
they are also examples of distributed processing. Such use presently 
involves mostly minicomputers. However, the availability of inexpensive 
microprocessors has opened entirely new vistas for distributed processing 
of this type. Because the technology is still very young, it has several 
obstacles to overcome before its promise will be fully realized. Com 
munication among multiple processors will probably turn out to be one 
of the main obstacles, particularly when the data transfer rates are 
high. [10j, [11] 
PHOTOGRAMMETRIC APPLICATIONS 
Photogrammetric equipment we are all familiar with contains a spec 
trum of digital elements. These range from simple encoders and counters 
of coordinate recorders to fairly sophisticated computer control of analy 
tical plotters. [12] Beyond that, advanced systems, such as the Epipolar 
Scan Stereomapper [13] extend the spectrum into high speed parallel pro 
cessing and operation of multiple processors in a real-time environment. 
As promised earlier, we will now explore the trends - realized or 
realizable - in the development of that spectrum.