×

You are using an outdated browser that does not fully support the intranda viewer.
As a result, some pages may not be displayed correctly.

We recommend you use one of the following browsers:

Full text

Title
Transactions of the Symposium on Photo Interpretation

WORKING GROUP 1
BIRNBAUM
We will present the findings for this photo-mode study. The prime question
is what difference, if any, is there in interpreter performance, that is attributable
to photo-mode, i.e., to whether the interpreter views negative transparencies,
positive transparencies, positive prints (non-stereo), or positive stereo prints.
There are three analyses that were conducted: 1. An analysis of accuracy
and completeness as a function of mode; 2. An analysis of confidence in right
and wrong responses as a function of mode; and 3. An analysis of accuracy
as a function of time separately for each mode. The sample consists of 64 photo
interpreters with an average of five years experience. Four sets of photographs
representing four different performance measures were used in this study.
Table 1 describes these measures. The content of two of these is such that the
desired identifications are individual objects. The scales of these photo sets
are 1 : 5,700 and 1 : 8,000 respectively. The content of the remaining two sets
consists of object complexes. The scale of both of these sets is 1 : 10,500.
Table 1. Description of performance measures
Each set of photographs consists of the same photograph reproduced in
four modes: positive transparency, negative transparency, positive stereo prints,
and positive non-stereo prints. Each of these was also reproduced at a lower
quality level thus providing two photo qualities and making possible an analysis
of mode-by-content-by-quality.
In order to conduct our studies we must be able to measure interpreter
performance. This requires the use of photographs the content of which is
known to the researcher but, of course, not to the photo interpreter test subject.
In addition, the photographs used and the requirements imposed on the inter
preter should be as realistic as possible. Also, the interpreter is provided with
relevant materials and information normally available to him, such as maps
and specific information about the deployment of friendly and enemy forces,
as well as equipment that is part of his photo interpretation kit. Essentially,
then, the performance measure constitutes a work sample with a known input
and with the opportunity for measuring interpreter output.
The determination of what constitutes the content of a given photograph
depends in all cases on the consensus of independent interpretations by expert
interpreters. In addition, available records were studied to verify the basic ana
lyses. In those rare instances when agreement could not be reached on the iden
tification of a particular object, the object was scored neither right nor wrong.
The performance indices used in these studies are a direct function of the
nature of the responses the interpreter makes. The interpreter can correctly
identify an object. The number of correct identifications (number of rights)
Measure
Content
Scale
1
Individual Objects (e.gvehicles)
1 : 5,700
2
Individual Objects (e.g., small structures)
1 : 8,000
3
Object Complexes (e.g., R.R. yards)
1 : 10,500
4
Object Complexes (e.g., fuel storage)
1 : 10,500