STEREOSCOPIC PERCEPTION WITH SINGLE PICTURES

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78B05171A000600070010-6
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 23
• Document Creation Date: December 28, 2016
• Document Release Date: April 4, 2003
• Sequence Number: 10
• Content Type: REPORT

Approved Eor Release 2003/05/14: CIA-RDP78B0 01A000600070010-6
.r.++mT,'1AT wTTR SINGLE PICTURES
STAT
ABSTRACT
Single photographic transparencies are shown
to produce stereoscopic perception of the recorded
scene when displayed with an optical system that
produces binocular image disparity. An optical
system producing differential illumination and
differential aberrations of the binocular images
is described.
The conditions leading to single picture
stereopsis are reviewed. The contribution of
factors like degree and nature of image disparity,
color, illumination, sharpness, orientation of
contours, depth perception cues, learning, to the-
observed stereoscopic effect is surveyed.
The proportion of pseudoscopic to true stereo-
scopic perception of the recorded image, the depth
information content of the single image and the
possibility to improve performance of optical
systems by using knowledge on how the visual
system handles information are discussed.
Declass Review by NIMA/DOD
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved F'Release 2003/05/14: CIA-RDP78B05 A000600070010-6
INTRODUCTION
it is well known and understood how to prod-
uce a stereoscopic visual simulation of a scene
by presenting to.the eyes a stereoscopic pair of
photographs. These pictures, taken from two pos-
itions laterally displaced relative to each other,
record most of'the retinal information that would
be registered by the eyes in a fixed position.
When presented to the eyes, such a stereogram
simulates the real scene and reproduces the stereo-
scopic perception of the original..
The presentation of two different images, one
for each eye, is so taken for granted as being the
only way to produce stereoscopic perception of a
recorded scene that it was rather surprising to
discover that it can be done also with single
images. In other words, not all the visual infor-
mation imparted by real objects in three-dimen-
ional space is necessary to produce stereoscopic
perception. It is the object of this paper to show
how, why and under what conditions stereopsis of
photographically recorded scenes can be synthesized
from monocular information.
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved FurRelease 2003/05/14: CIA-RDP78B051,A000600070010-6
The phenomenon could find application in
display and simulation devices, in visual
enhancement for detection systems and for plain
home slide and.movie viewing.
Besides the possible applications (that
frequently don't come out to be those envisioned
by an author), stereoscopic perception of single
images is valuable for the fresh insight it offers
into the process of stereopsis.
OPTICAL DISPLAY SYSTEM
The main condition'for stereoscopic percep-
tion of single pictures. is an optical system
capable of artificially producing ocular image
disparity. One such system found to be very
effective is provided by a good quality slide
viewer. Its observation section is a large two-
lens magnifier of an aperture large enough to
.contain both eyes, and its illumination section
is a diffusing screen illuminated by a concen-
trated light source (incandescent bulb)-(Fig. 1).
The system produces two types of disparity:
luminance disparity and geometric disparity.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved FbwRelease 2003/05/14: CIA-RDP78B051 4000600070010-6
Any element of the image in the transparency will
have a different luminance for each eye because
of the anisotropy of the illumination produced by
the concentrated source and the diffusing screen.
Image elements in the left half of the screen will
produce a retinal image of higher illuminance in
the left eye and a lower illuminance in the right
eye and vice versa. If the concentrated light
source is shaped as a vertical line, the left
right luminous disparity will be uniform over the
whole height of the picture. A partially diffusing
screen (like ground glass rather than opaline plas-
tic) will increase luminous disparity. Luminance diaparity
will decrease with distance from light source to
screen, the extreme case being collimated light.
Because the optical paths from the image
element to each eye are not symmetrical, the
aberrations of the optical system will not be the
same for the left eye and the right eye image of
the same picture element. Thus, the two retinal
images of the'picture elements possess geometric
disparity.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved For F ase 2003/05/14: CIA-RDP78B051714 &0600070010-6
For a symmetric position of the eyes with respect
to the axis of the system, image elements situated
around the vertical median line of the image do not
exhibit any disparity. Due to the integral charac-
ter of the stereoscopic vision process, however,
there is no discontinuity in the stereoscopic aspect
of the scene. Various other optical systems for
producing retinal image disparity of the same trans-
parency are conceivable, but the one described
appears to be the simplest and proved effective in
detecting and demonstrating single picture stereopsis.
DISPARITY AND COLOR IN STEREOPSIS
Disparity and color are factors independent
on information. They produce stereopsis.without
any previous training or knowledge from the part of
the observer. In real life (by "real life" is meant
the experience encountered during the evolutionary
process that led to human vision as it is today)
disparity of the retinal images of the same object
was always produced by the position of the object
relative to the eyes. This is why the sense of
vision always interprets disparity as produced by
relative position in space and "disparity alone
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved Fq,Welease 2003/05/14: CIA-RDP78B051A000600070010-6
provides the necessary and sufficient stimulus (l)
for the emergence of the stereoscopic experience".
A survey of past research on space perception
and stereopsis discloses a variety of findings that
explains completely the phenomenon of single picture
stereopsis and accounts for all the factors involved.
Luminance disparity, geometric disparity and color
stereopsis were shown to produce stereopsis.
Luminance disparity was produced by a neutral
density or a colored filter in front of one eye.
In the Puifrich stereophenomenon(2) the bob
of a pendulum moving in a frontal plane seems to
describe an elliptical path, nearer to the observer
when moving in one direction, and farther from him
when returning. Changing the filter to the other
eye reverses the effect. The magnitude of this
stereoscopic effect was found to increase with
filter density.
The same stereoscopic effect can be obtained
substituting the neutral density filter with illum-
ination of one eye (using an "additive filter" as
it were)(3'.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved Fb elease 2003/05/14: CIA-RDP78BO517JA000600070010-6
In irradiation stereoscopy (4,5) two illumin-
ated white rectangular sheet-like objects in a
frontal plane with a black background, when observed
with a neutral density filter appear to have rotated
each by a small angle about its vertical axis of
symmetry.
Stereoscopic depth and stereoscopic acuity are
enhanced by use of a colored filter in front of one
(6)
eye
Geometric disparity with artificial objects was
used to investigate stereopsis(7).
The objects are random dot patterns, where the.
difference between the left and right eye figure
consists in the displacement of patches of dots in
the two otherwise identical figures. Stereoscopic
perception of the displaced part of the figure is
produced. The stereoscopic effect is so strong,
that a wide range of optical illusions can be re-
produced with the "stereo patches" of the random
dot figures
The production of stereoscopy by geometric
disparity is well known in astronomy, where it is
used for rapid identification of changes in the
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved W Release 2003/05/14: CIA-RDP78B051A000600070010-6
map of the sky. Two photographs of the same region
of the sky, taken at different periods, are observed
in a stereoscope. Any new object like a supernova
or comet, or an object that changed its place
between exposures, will appear as floating in front
or behind the plane of the picture.
The same technique is used in the observation
of aerial intelligence photographs to detect changes
in a scene. It was observe3that geometric disparity
by
produced/observing a plane slide with a larger magni-
fier containing both eyes makes the slide appear to
be convex(9).
Color stereoscopy is the second experience
independent condition that produces stereoscopic
perception of single images.. Small luminous objects
of different colors appear to be at unequal distances
(10)
. Colors in the long wavelength part of the
spectrum appear usually nearer than the shorter
wavelength colors.
The color stereoscopic effect depends also upon
(12)
the relative luminosities .
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved Release 2003/05/14: CIA-RDP78BOW1A000600070010-6
The color stereoscopic effect is rather striking
and, depending upon other information present in the
image, tends to either enhance the true stereoscopic
perception or to create pseudoscopic effects, where
details do not occupy their natural place in stereo-
scopic space.
"REAL WORLD" SCENES AND DISPARITY
In accordance with sound scientific method,
simple elements of the phenomena were isolated for
study and this iswhy previous research on stereopsis
deals with simple artificial objects. This approach,
albeit the only practically possible, does not always
handle satisfactorily the complexity of the real
world. A case in point is the visual'process where
previous experience of the observer plays such an
important part.
Inducing binocular disparity for complex objects
like photographic, colored and black and white,trans-
parencies of real life scenes will put in evidence
the interdependence of the factors involved in
stereopsis.
Artificial disparity will always produce
some stereoscopic effect, that will be supplemented
by the rest of the depth information present in the
picture to produce stereopsis.
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved For Release 2003105/14: CIA-RDP78B05 AA000600070010-6
Name
cases, this stereoscopic effect
In many
pairs-of stereograms,
is akin to that produced by
bserver sees a stereoscopic simulation of the
The o
photographed scene, rather than some stereos in itself.
p The
distortion of the transparency as an object/
details of objects in the
recorded scene appear in their right place in
stereoscopic space, as they would with pairs of
stereograms.
effect thus produced varies in
The stereoscopic
intensity and realism, depending on a number of
conditions that will be discussed further.
Nineteen typical slides were shown to 37
observers of ages ranging from 25 to 62. All but
one of the subjects described the same stereoscopic
effects. One subject, who disclosed uncorrected
eyesight, different in both eyes, did not discern any
stereopsis.
STEREOSCOPIC PERCEPTION
PRODUCTION OF SINGLE PICTURE
Luminous and geometric disparity, as well as
where shown to produce stereoscopic
in the literature
color stereoscopy
erception. There is no indication however/of any
p tion.
factor that would cause true stereoscopic percep
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved For.Release 2003/05/14: CIA-RDP78B051r000600070010-6
The single image does not contain stereoscopic infor-
mation, like true geometric disparity for the left-
right eye images of objects, or true luminosity
disparity. The stereoscopic perception of the image
is produced using artifically produced geometric and
luminous disparity. Nevertheless, the scene is
perceived in true stereoscopic perspective.
This is not violating any conservation law. We are
not getting something for nothing.
In the phenomenon of space perception, the
stereoscopic cues represent only a small part of the
wealth of visual information, present and stored,
used by the brain to construct the sensation of
visual space. It appears that the rest of the visual
information, the depth perception cues combined with
artificially induced stereoscopy with its enhancing
factors are sufficient to produce a valid simulation
of visual space using a single picture. The more
cues, the more intense and realistic this perception will be.
A study of stereoscopic effe.ctsshown by a number
of photographic slides leads to the conclusion that
the depth perception cues and the stereoscopy enhancement
factors are contributing statistically to the
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved*,pr Release 2003/05/14: CIA-RDP78B04W1A000600070010-6
stereoscopic effect. Below a certain minimum, all
there is left are stereoscopic illusions, with red or
orange luminous spots in the image "floating" in the back or in
front of it. In the presence of enough cues, at the
other extreme, we see a stereoscopic image that could
hardly be told from one produced by a conventional
stereogram.
The depth cues and the stereopsis enhancement
factors are not independent parameters. They rein-
force and complement each other in various degrees,
according very much to the statistical conditions
prevalent in the visual world of the observer.
STEREOPSIS ENHANCEMENT FACTORS
These factors are connected with*the technical
characteristics of the picture.
Sharpness of focus and contrast were recognised
as factors of stereoscopic acuity (13)
Vertical contours are essential in stereoscopic
vision(14). The more vertical contours in a scene,
the more marked the stereoscopic effect. Also,
stereoscopic acuity is higher for rods than for
points.
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved fFpr Release 2003/05/14: CIA-RDP78BO W1A000600070010-6
High retinal illumination increases stereoscopic
acuity (15)(16) It should be noted at this point
in
that stereopsis is present both/photopic (cone) and
scotopic (rod) vision.
It can be seen that these technical parameters as
well as others to be discussed further
/coincide largely with the desiderata for any good
photograph.
DEPTH CUES AND INFORMATION
The accumulation of monocular depth perception
cues determines the degree of realism in.single
picture stereopsis, since these are the information
that characterizes the cognitive part of the visual
process.
Most depth cues should be called'familiarity
cues, since they are acquired through a learning
process. It is known that depth perception is
impaired in unknown surroundings and the dependence
of depth cues on cultural background was demon-
strated(17,18,19)
Monocular depth perception cues
contribute to the production of single picture
stereopsis on a cumulative basis and can be classified
into(20): interposition; size of known objects;
association with other objects; perspective of
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 SW-or Release 2003/05/14: CIA-RDP78B0 T1A000600070010-6
parallel lines; vertical position in field; fore-
shortening due to perspective; distribution of light
and shadow; aerial perspective*
lt of matching the
All these cues are the resu
details of an observed scene with the model of the
world existent in the observer's mind
We remarked already that in single picture
s the quantity of information imparted by
stereopsi
the cues must be above a certain threshold. In other
words, single, too few or too weak cues will produce
an ambiguous, unstable or throughout pseudoscopic
car taillights in a dark scene
erception. (E.g.,
P
will appear floating in stereoscopic space,
disconnected to the cars themselves.) The presence of familiarity cues in a single
picture is essential for the fidelity of stereoscopic
picture
perception. We observed that in all our subjects,
stereopsis was the more positive the more explicit
slides, and the more luminous and sharper the
the
picture. For dark or inexplicit details or back- real life
In
ground, stereopsis was diminished or retarded.
conditions, stereopsis depends heavily on secondary
cues. Ambiguous or misleading cues were used to
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved F,ap Release 2003/05/14: CIA-RDP78B054 1A000600070010-6
create spatial illusions that were not dissipated by
binocular observation. The stereoscopic localization
of scene elements was falsified
In case of ambiguous visual cues, the observer
can even be.conditioned at what depth to perceive the
object by conditioning with acoustical stimuli(22).
STEREOPSIS OF COLORED SLIDES
Taking any batch of colored slides representing
familiar scenes or subjects and classifying them by
the degree of stereopsis, it was observed that
the most effective were always the ones on which
visual information is clear and obvious. Not so
surprisingly, these are the ones that are good from a
photographic standpoint also. They ate in good focus,
there are no overexposed or underexposed details, the
elements of the scene are obvious, uncluttered, well
detached visually from one another. They have good
color contrast and are not ambiguous. The objects
present texture; that means there are no large areas
unbroken by detail. Or, to better stress the infor-
mation aspect, the "message" of the picture is clear.
It is worth remarking that good conventional stereo-
grams present the same characteristics. In case that
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved`? pr Release 2003/05/14: CIA-RDP78B04ir71A000600070010-6
only part of the picture was up to these standards,
then only this part induced stereopsis. The rest
either did not induce stereopsis, or produce d
pseudoscopic effects of false perspective, or stereo-
scopic illusions of objects "floating" in visual
space without connection to the scene recorded in the
picture.
According to the photographic quality of the
picture and to the abundance of cues, weak or strong
stereopsis was experienced. For unfamiliar
scenes, stereopsis
may appear only after a few
seconds, when the objects are identified.
Stereopsis with single pictures could facilitate
the study of various factors of visual'perception.
The use of photographs makes the manipulation of the
variables more flexible(17).
CONCLUSION
Stereoscopic perception can be reconstituted
from artificial image disparity and real depth cues.
The intensity and fidelity of the phenomenon is
dependent on the quantity of information conveyed by
the picture.
Stereoscopic perception with single pictures
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved Fo lease 2003/05/14: CIA-RDP78B0511000600070010-6
stresses the nature of the process of vision as part
of a highly integrated and complex information acqui-
sition and processing system. (How complex it is can
be inferred from the fact that of all senses it is
allocated the largest area on the cerebral cortex.)
Vision is more than a sum of image acquisition and
processing mechanisms coupled to a computer that
interprets the images. We have there a learning
system that is working in real time, continuously
adjusting and extending its internal model of the
world and comparing to it any inputs.
On the more practical side, single picture
stereopsis can be used for image enhancement as
approached from the visual information acquisition
and processing standpoint rather than by what we
elect to call "brute force," i.e., enhancement of the.
energy related parameters like illumination, power,
.intensity.
Retinal image disparity can be created and the
effects. of color can be used to intensify perception
and discrimination of details in a picture.
Besides its sensitivity to the energy carried by
electromagnetic radiation, human vision possesses
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approvedr Release 2003/05/14: CIA-RDP78B05,71A000600070010-6
sensitivity to information. This "information sensi-
tivity" is much harder to define quantitatively, but
is nevertheless there and makes itself obvious in
occurrences like the fast identification of spots
presenting ocular disparity in image pairs. Visual
information sensitivity, if adequately put to use,
could improve the performance of many visual systems
without need to push the state of the art in their
energy collection, transfer and amplification capa-
bility.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approvedffpr Release 2003/05/14: CIA-RDP78B0W1A000600070010-6
REFERENCES
1. K.N. Ogle, The Optical Space Sense (The Eye, ed.
by H. Dawson, Academic Press, New York, 1962) p.
387,
2. C. Pulfrichi Naturwissenschaften, 10, 553 (1922)
3. A.L. Diamond, J. Opt. Soc. Am., 48, 887 (1958)
4. C. Munster, Z. Sinnesphysiol., 69, 245 (1941)
5. P.A. Cibis and H. Haber. J. Opt. Soc. Am., 41,
676 (1951)
6. H. Schafer and F. Ebner, Pflug. Arch. ges.
Physiol, 249, 637 (1948)
7. B. Julesz, Science, 145, 356 (1964)
8. B. Julesz, "Foundations of Cyclopean Perception"
paper presented at the 1969 Fall Meeting of.the
Opt. Soc. Am.
9. R. Kingslake, (Applied Optics and Optical Engin-
eering - Academic Press, 1965), Vol. I, Ch. 6, p.
235
10. W. Einthoven,
Brain, 16, 191 (1893).
11. J.J. Vos, J. Opt. Soc. Am., 50, 785 (1960)
12. G.E. Mount, H.W. Case, J.W. Sanderson and R.
Brenner, J. Gen. Psychol. 55, 207 (1956)
13. R.G. Frey, Wien. Med. Wschr. 103, 436 (1953)
14. E.E. Andersen and F.W. Weymouth, Am. J. Physiol,
64, 561 (1923)
15. C.G. Mueller and V.V. Lloyd, Proc. Natl. Acad.
Sci., U.S. 34, 223 (1948)
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 Approved `Fwr Release 2003/05/14: CIA-RDP78B0Ay71A000600070010-6
16. A. Litt Am. J. Psychol., 62, 159 (1949)
17. H. Leibowitz, R. Brislin, L. Perlmutter, R.
Hennessy, Science, 166, 1174 (1969)
18. M.H. Segall, D.T. Campbell, M.J. Herskovits,
Science, 139, 769 (1963)
19. M.H. Segall, D.T. Campbell, M.J. Herskovits,
The Influence of Culture on Visual Perception
(Bobbs-Merrill, Indianapolis, 1966)
20. K.N. Ogle. "Researches in Binocular Vision"
(Saunders, Philadelphia, 1950) 50
21. W.C. Gogel, J. Exp. Phsychol., 47, 335 (1954)
22. B. Julesz. "Acoustical Conditioning of Binocular
Depth Perception" paper presented at the 1964
Spring Meeting of the Opt. Soc. Am.
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved F Release 2003/05/14: CIA-RDP78B0SU4A000600070010-6
Figure 1
Optical system producing luminous and geometric disparity
with a single transparency.
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070010-6
 Approved K Release 2003/05/14: CIA-RDP78B054A000600070010-6
W
w Q H
~0
F- 0 (
~
Jo
0
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070010-6
 044w- U504,0- 6 0-14
--~ ^,o- 4"04.1"
-~''~1.GG.. /L~,~c.ri+~?w- 2 7` ) .
STAT
Approved For Release 200P/05/14 : CIA-RDP78BO5171A000600070010-6