retric
ceters
attempts
1 the
ages
zo
EL, 1952)
xhibited
1d
"Camera
Fr a
right
these
rent
iS
the
> and
Eerans-
oric
Cy
rations
at
aphic
Ary
raphy ;
on
che
quality
f two
res
of the visual acuity of the image analyst which are almost
identical with these three image quality characteristics.
Specifically, it is highly relevant for three corresponding
questions to be asked: (1) to what extent can any given image
analyst (or applicant for work involving image analysis) per-
ceive differences in tone or color contrast between an image
and its background (e.g., does he have some kind of color
blindness?); (2) to what extent can he resolve fine detail
and thus exploit the sharpness of an image?; and (3) to what
extent can he perceive stereoscopic parallax, if at all?;
unfortunately some image analysts have only one good eye.
b) It follows from the above that a competent image analyst, on
examining a given type of imagery, can determine to what
extent that imagery is of suitable quality in terms of each
of these three attributes. If, on making such a determination,
he finds deficiencies in one or more of these attributes and
wishes to make improvements in that attribute when acquiring
additional imagery in the future, he needs only to turn to
the list of factors described above that govern the quality
of that attribute. In so doing he can promptly learn what
approach needs to be used in rectifying the deficiency, what-
ever it may be, when more photography is being taken.
Two measures of image quality that were being developed at about
the same time as the above have become known, respectively, as
"acutance" and the "modulation transfer function". Both of them
probably are more precisely measurable and consequently more
appealing to the physicist and mathematician than the "tone-
sharpness-parallax" concept that has just been described. But
experience has shown that both of them are much more difficult
for the average image analyst to comprehend, and to identify
with. Furthermore, neither acutance nor the modulation transfer
function gives due consideration to stereoscopic parallax. In-
stead, each concerns itself primarily with the "sinusoidal res-
ponse curve" that is obtained when a plot is made of the change
in tone or brightness per unit of distance along the edge or
perimeter of a feature, as imaged on photography. Acutance deals
with only one such edge (e.g., the classic photograph of a "knife
edge") while the modulation transfer function deals with a whole
sequence of such edges, repetitively spaced, but at progressively
closer distances, as in a resolution target.
By about 1960, a systematic approach to the factors governing the
amount of information derivable from image analysis tended to
deal in one way or another with four components, only the first
of which has been alluded to in the preceding paragraphs. These
components, at least as set forth in one representative treatment
(American Society of Photogrammetry, 1960) were described in the
following statement: "If aerial photographic interpretation in
any of its applied forms it to be employed successfully, four
conditions must be satisfied: (a) The aerial photography must
provide images of suitable quality for extracting the type of
information that is to be obtained through photo interpretation;
(b) the men performing the photo interpretation work must have
been properly selected and trained; (c) the equipment used in
viewing, measuring, and interpreting the photographic images
must be of suitable quality; and (d) the methods and techniques