better 
erable 
o that 
in the 
oot is 
horter 
2 
e zero 
Thus 
speed 
rvo is 
se, the 
cannot 
a con- 
‘vor is 
of the 
f slow 
; these 
er, the 
speeds. 
> com- 
in the 
ilar to 
id. will 
by the 
of this 
instru- 
e com- 
ay at- 
s used. 
in dis- 
r is in 
to cor- 
light, 
ber of 
able to 
opera- 
d more 
> input 
shed in 
om are 
  
  
THE DESIGN OF PHOTOGRAMMETRIC PLOTTERS, HELAVA 
well known, a complete mathematical analysis of the performance of the entire system is 
possible. However, in a plotter a human operator is employed, and in spite of the im- 
portance of the human element in many control operations the performance of a human 
link in a control system is not too well known. About the only things that are known are 
that his response is relatively slow, he tends to “overshoot” and he is inclined to proceed 
step by step in his controlling action. Up to a point the human operator may be replaced 
by image correlation devices. Their performance is perhaps easier to analyze and to state 
in definite terms, but too little experience is available so far to be sufficient for a detailed 
analysis of the performance of a plotting system. : 
Although we lack detailed information on some basic elements, it is still possible to 
study the system in general terms. The response times of different parts of the system 
are of particular interest for accuracy and stability reasons. The accuracy is obviously 
affected if the response time of the “plotter” is long as compared to that of the “opera- 
tor”. In this case the output of the plotter part of the system could not follow the changes 
of its input closely enough, and would fall behind. This may develop to an instability 
condition if the input varies in an oscillating manner. An increasing frequency of the 
input oscillation causes the output to fall behind more and more until the phase error 
becomes about 180°. If, at the same time, the output: input magnitude ratio is about one, 
instability is encountered. For these reasons we should strive towards a solution where 
the response time of the plotter is small as compared to that of the operator, and where 
the output : input ratio is small. This means that the computer-servomechanism combi- 
nation must be fast and the motions to be controlled by servomechanisms small. 
There is no doubt that a highly satisfactory plotter can be designed and realized for 
the human operator. The computer-servomechanism combination can be considerably 
faster than what is required from either an accuracy or stability point of view. This 
situation becomes less clear when a mechanical operator is considered. So far the speeds 
of demonstrated image correlators do not greatly exceed the speed of a human operator, 
but these devices are inherently capable of operating at much higher speeds. When the 
higher speed is realized, the requirements to be set to the response time of the plotter 
part of the system become correspondingly severe. We know today about computer im- 
provements that permit the computing speed to be increased ten to hundred fold. Com- 
parable improvements may be found for the performance of the servomechanisms by 
processing all the information on a stereopair by changing two of the computer input 
variables (usually X and Y) in a regular manner while only one of them (usually Z) 
varies in an oscillating manner. Eventually the mechanical character of the motions con- 
trolled by the servomechanisms will be the limiting factor, because the oscillating input 
'ariable is not necessarily associated with a corresponding mechanical motion and there- 
fore may be very rapid. Then for higher speeds the servoing should be accomplished by 
employing massless elements, such as the flying spot of a cathode-ray tube. 
4. Design considerations. 
In a photogrammetric mapping problem three sets of coordinates are of primary 
importance. These are the ground coordinates, or coordinates of the photographed object, 
that we denote here X, Y, and Z, and the plane coordinates on two photographs (x, ¥;; 
a, V9). The basic problem in a plotter is to correlate these points and to collimate the 
homologous points. The correlation and collimation should be correct within the obser- 
vation acuity, and the plotting should be possible no matter how complicated the relation 
between the primary coordinates. This is the primary task, and it should be kept in mind. 
There are also other coordinates often referred to, notably the instrumental coor- 
dinates (x, y, z) and the plotting table coordinates («', y’, 2’). These represent inter- 
mediate results and have been accepted for various reasons — because the relationship 
between the coordinates of the photographs and the final map coordinates has not been