62
SYMPOSIUM PHOTO INTERPRETATION, DELFT 1962
therefore is one measure of interpreter performance. The interpreter can mis-
identify objects. Number of wrongs, therefore, is another measure of his per
formance. Further, he can fail to respond where he should. Number of omits
thus constitutes a third measure. In addition, two derivative measures were
developed, one accuracy, the other, completeness. Accuracy is the proportion
of correct to total responses an interpreter makes; it is the ratio of number of
rights to number rights plus wrongs
R
R+W
Completeness is the proportion
of correct responses to total extractable information that is in fact extracted
from a photograph; it is the ratio of number of rights to number of rights plus
omits
R
R+Ö
Accuracy and completeness as a function of mode
The accuracy and completeness analysis was conducted separately for the
two photo qualities, high and low. Within each quality four performance
measures were analyzed separately, accuracy after six minutes of work, accu
racy after thirty minutes of work and similarly completeness after six and
thirty minutes of work. Table 2 shows the accuracy results for the four
different modes of high photo quality. The comparison here is for each row.
None of the differences is significant. Table 3 presents completeness results
for the four different modes at high photo quality. Again the differences
Table 2. Percent accuracy by photo mode for high photo quality
Positive
Negative
Stereo
Non-Stereo
Accuracy
Transparency
Transparency
Print
Print
Initial
53
53
54
47
Final
42
38
44
40
Mode differences are not significant.
Table 3.
Percent completeness by photo mode for high photo quality
Positive
Negative
Stereo
Non-Stereo
Completeness
Transparency
Transparency
Print
Print
Initial
14
10
9
10
Final
33
27
31
30
Mode differences are not significant.
Table 4.
Percent accuracy by photo mode for low photo quality
Positive
Negative
Stereo
Non-Stereo
Accuracy
Transparency
Transparency
Print
Print
Initial
40
41
52
39
Final
30
33
39
37
Mode differences are not significant.