WORKING GROUP 1 
BARRETT 
55 
Table 3. Problems and prospects 
Problem Volume of Raw Data Input (Photos) 
User’s need Some Level of Auto-Screening 
Typical research approach Auto Pattern Matching Devices 
Major difficulty Impossibility of predicting all possible pattern variations 
SDC approach Self-Learning Pattern Recognition, Self-Generated and 
Dynamically Changing and Improving Master “Patterns” 
We are not expecting that an automatic photo interpretation machine is 
suddenly going to burst forth from our research, or from any other similar 
research of which I have any knowledge. Personally, I cannot yet conceive 
of such a device - of a machine equalling the capabilities of a human in 
detailed photo analysis - but I do visualize a machine which can do very 
rapid course screening of large amounts of photographs to determine the 
presence or absence of certain gross features expected on the photography. 
How far this capability may ultimately progress, I do not know. At the present 
time, our’s is a research effort and certainly not a system design program. 
Most present-day computers require that their input be in digital form, and 
there are a number of devices available for so quantizing photographs [12, 13]. 
We have concentrated on processing the photos in the computer after the 
initial digital input operation. Research-type test operations are now being 
Fig. 1. 
Operator function 
OPERATOR 
1 
1 
1 
1 
0 
1 
0 
OPERATOR 
1 __ 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
HI 
TA 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
0 
U- 
UJ 
Q, 
JAT 
OR 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
o 
z 
1 
HI 
T E 
1 
1 
£ 
. 
1 
1 
1 
- 
cc 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1 
1