STEREOSCOPIC VISION APPLIED TO PHOTOGRAPHY

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78B05171A000600070052-0
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 21
• Document Creation Date: December 28, 2016
• Document Release Date: April 7, 2003
• Sequence Number: 52
• Content Type: OPEN

Approv~ or Release 2003/05/14: CIA-RDP78f171A000600070052-0
STEREOSCOPIC VISION APPLIED
TO PHOTOGRAMMETRY
By PROFESSOR W. D. WRIGHT
Imperial College of Science
Meeting of the Society, 4th December, 1954.
ABSTRACT
In Part I the visual processes involved in depth perception are briefly
described and discussed, while in Part II their application in stereo plotting and
allied observations is considered. Reference is made to the visual axis, the need
for and nature of eye movements, the conception of the horopter space within
which fusion of the binocular images occurs and objects are seen in three-
dimensions, and the tolerances for fusion implied by the existence of Panum's
areas. Monocular clues make an important contribution to the perception of
depth, and the apparent location of an object in space is determined by the
factors of retinal disparity, convergence, angular size, perspective, motion
parallax, elevation, aerial haze, etc. The following problems are then discussed:
The significance of the horopter in relation to the general contour in a stereo
photograph; the fusion of the images in a stereo plotting machihe; the arguments
for and against some convergence of the eyepiece axes; the special problems of
spectacle wearers; the functions of field stop. An analysis of the visual task in
contour plotting is attempted, but a more complete analysis would be possible
if records were available of the observer's eye'moveritents while` plotting. Finally,
chromatic aberration in relation to the Multiplex system is considered, and
also in connection with the stereoscopic effect observed with differently coloured
objects lying in a plane.
The Optical System of the Eye; the Visual Axis; Eye Movements
Fig. .1 shows a horizontal section through the human eye and illustrates
its refracting system comprising the cornea and crystalline lens, the ciliary
processes controlling the focusing power of the lens, the iris which controls the
cone of light focused by the lens, and the light-sensitive membrane, the retina,
lining the inner wall of the eyeball and on which the image is focused. The
detailed structure of the eye is dealt with exhaustively in text-books on physio-
logical optics and need not detain us here, (See, for example, Duke-Elder,
Text-book. of Ophthalmology, Vol. I, 1932.)
For our purpose it is of interest to contrast the eye with three other types
of optical device-the telescope, the camera, and the sky lens. The telescope
Approved For Release 2003/05/14f~1~r1~052-0
 Approv or Re lease. 20.0.3/05114.. CJA-RDP_Z8.6,01Z1AQ00K0070052-0
Fig. 1. A horizontal section of the human eye.*
can be used to study the fine detail of a distant object, but as it normally has
a rather small field of view of perhaps only a few minutes of arc, it has to be
deliberately directed to each point in the scene which it is desired to examine
in detail. On the other hand, the average camera covers a much wider field
of, say, 50?, and both the definition of the image and the distribution of the
light-sensitive silver halide grains are essentially uniform over the area of the
plate or film. The sky lens is a specially designed optical system for photo-
graphing the clouds in the sky, and receives light from very nearly the whole
hemisphere of the sky in one photograph.
The eye is rather remarkable in sharing some of the properties of both
the sky lens and the telescope. The lens and retina are capable of collecting
light from a field of slightly more than 180 ?, and it is only the obstructions of
the face-nose, cheek, forehead, etc.-which reduces this angle of acceptance
for a single eye. The two eyes together do,.in fact, cover a field slightly in
excess of 180? in the horizontal plane. On the other hand, if the fine detail
of any point in the scene is to be examined, then the observer has to direct
his eyes to the point under regard. The resolving power of the retina has its
maximum at the' fovea (see Fig. 1) and a line through the foveal centre and the
nodal point defines the visual axis of the eye (Fig. 11), which has to be directed
to the point of interest at any given moment. The details of the whole scene
are then only perceived as a result of the rapid movements of the eye darting
from one point of the scene to another.
This pointer-like property of the eye is well brought out by measurements
of the resolving power of the eye at various angles from the visual axis leading
to the curve shown in Fig. 111. The peripheral retina does, of course, con-
'(From Schafer (3) Quain's Anatomy, vol. III, Pt. III. 10th Ed.Longmans, Green and Co., London 1894.)
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approvror Release 2003/05/14: CIA-RDP786171A000600070052-0
tribute to the perception of the general pattern of the scene, and it also has
other very important functions, such as enhanced sensitivity to movement,
and high light-sensitivity when the eye is dark-adapted and so on.
60 40 70 0 20 40
Angular Distance from Foveal Centre (degrees)
Fig. III. The variation of the visual acuity
across the retina.
? The movements of the eyes necessitated by the existence of the visual
axis are themselves of great interest and in all probability of fundamental
importance to visual perception. There are, in fact, two or three distinct
types of movement: there are, first, the small movements which occur even
when a small fixation target is being viewed as steadily as possible and which
imply a limited precision to the location of the visual axis; then there are the
saccadic movements in which the eye moves in a series of jumps and pauses;
such movements occur not only when viewing a scene but in reading, for the
eyes make a number of fixation pauses along each line of print. Lastly, there
are the pursuit movements in which the eye makes a relatively smooth con-
tinuous movement when following a moving target. Examples of these three
types of movement are shown in Fig. IV as recorded by a corneal reflex method.
(Lord, 1948; Lord and Wright, 1948, 1949, 1950.)
Binocular Vision
One obvious advantage of having two eyes is the increase in the field of
view. Fig. V shows that while some 120' of the central field is covered by
both eyes, the last 300 on either side of the peripheral field in the horizontal
plane is observed monocularly. In this diagram it is assumed that attention
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approver Rel
ease 2003/05/14: CIA-RDP78BQ,W'71A000600070052-0
71ME(.5 )
TIME ( stc )
(a) Movements during fixation on a target when
the observer was unable to maintain very
steady fixation.
(b) Movements during fixation on a target when
the observer was able to maintain very steady
fixation. (The effect ofsmall head movements
are also included in this record.)
(c) Pursuit eye movements while the observer.
was following a target attached to a swinging
pendulum.
(d) Saccadic eye movements recorded when the
observer was attempting a slow, continuous
sweep movement between two targets without
any moving target to follow.
Time is recorded horizontally in 1/Sthssecond.
The scale of the angular deflection is indicated
by the short vertical line to the left of each
record.
Fig. IV. Eye morement traces as drawn
from records by Lord and Wright obtained
0 0 corneal reflex method.
Fig. V. The angular fields of view of the
right and left eyes. (The scale of
the field is in degrees.)
is being directed to a particular point in a scene (represented by the zero point
in the diagram), and convergence of the visual axes on to this point is necessary
if double vision (diplopia) is to be avoided. The two foveic on which the point
is then imaged are in fact important examples of what are known as corre-
sponding points.
We can imagine that for each point on one retina there is a corresponding
point on the retina of the other eye such that, when both points are stimulated,
they give rise to a single fused image in the brain. This conception, as we
.shall see in a moment, may need to be modified, but on the assumption that
corresponding points in the two retinie are equally spaced from their respective
fovea, Muller's horopter, Fig. VI, can be constructed to give the locus in space
of points which will all appear single although viewed by the two eyes. This
theoretical horopter is a circle passing through the nodal points of the two eyes
and the point under foveal fixation.
Experimental determinations of the horopter are not in agreement with
this theoretical curve, and the particular shape that is recorded is a function
of the viewing conditions and also to some extent of the concept which is held
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved r Release 2003/05/14: CIA-RDP78BO 1A000600070052-0
Fig. VI. Miller's hypothetical horopter,
as definer! by the locus of points appearing
single ,fin binocular vision when P is
fixated, and assuming that corresponding
points in the two retinae are equally
spaced fi-onr their respective fovea!,
that is, L,M, = L,M,.
-1i 4
II---I-- a I a.I- J
mm-2
-,,
0
mm.4
-B
-12
10
C. 0
-10
-20
Fig. VIi. Experimental determination of the
horopter defined as the apparent frontal plane,
when a central point is being fixated
(after Ames, Ogle and Gliddon).
Distance of fixation point:
(a) 20 cm. (b) 76 cur. (c) 2 in.
Displacement front frontal plane plotted
against angle across field.
to define the horopter. Thus, Fig. VII shows the determination of the horopter
regarded as the locus of points lying in a frontal plane perpendicular to the
direction of view as measured for three viewing distances and with fixation
maintained on a central target in each case. (Ames, Ogle and Gliddon, 1932.)
Fortunately, the direction of the visual axes of the two eyes on to the point
of regard does not have to be exact to ensure single vision of the point. Fusion
will also be possible for other parts of the visual field provided the images
fall near to the corresponding points to within the so-called Panum's area
(which is in effect a kind of tolerance area). The size of the Panum areas
increases with the displacement across the retina from the fovea, but has an
angular diameter of the order of a few minutes of are at a degree or two from
the fovea. Walls (1952) has recently emphasised that the existence of the
areas leads to a different concept of the horopter from that of a surface; it
should rather be thought of as a space within which all objects will be seen
fused and in three dimensions. Objects lying in front or behind the horopter
space will appear double, and the horopter space might well be defined as
the space outside which objects do appear in diplopia. We should then
understand the perception of an extended scene in three dimensions as derived
from successive binocular fixations on different points in the scene, at each of
which some part of the scene is seen in depth.
Before considering the nature of stereoscopic vision in more detail, we
should note the remarkable co-ordination required of the extra-ocular muscles
if the orientation of the two eyes is to be correctly maintained during the very
rapid fixation movements that occur, even when account is taken of the toler-
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved ftf Release 2003/05/14: CIA-RDP78B01AD00600070052-0
ances allowable on account of Panum's areas. Fig. VIII illustrates the anatomy
of the eyes in their orbits and of the extra-ocular muscles which control the
eye movements. The degree of co-ordination that is required, not only between
the two eyes but between the eyes and the head, is almost incredible, but is
nevertheless successfully achieved in persons with a normal oculo-motor
system and normal muscle balance.
Fig. VIII. Diagram showing the eyes in their
orbits and the extra-ocular muscles.
The Perception of Depth
The primary clue to distance arises from the binocular parallax associated
with the different view-points of the two eyes. Each eye forms a slightly
different image of a three-dimensional scene and the retinal disparity that
exists for the very simple case of the two object points A and B at different
distances from the observer is illustrated in Fig. IX. If this disparity BLBR
is not excessive the brain will be able to fuse the two images by processes which
are only imperfectly understood but which evoke a sense of depth, so that
A and B are clearly seen to be at different distances. From what has been
said above, it will be recognised that this fusion will occur provided A and B
lie within the same horopter space. However, experiment shows that some
sense of depth is retained even in diplopia (Wright, 1951). Fusion is not,
therefore, an essential element in depth perception, although it must be an
important contributory factor.
A normal scene will be vastly more complicated than Fig. IX and will
in general involve many objects at various distances from the observer, hence
as the scene is scanned by the observer some of the objects will be seen at a
given moment fused and in three-dimensions, while others will be seen in
diplopia. Then as the fixation is changed, a new part of the field will be seen
fused and so on. Just how the relative locations in space of the successive
fixation areas are interlocked, is still a matter of dispute, but experiments
suggest that the varying convergence of the two eyes as a near or distant object
is being viewed, may be an important link (Wright, 1951).
When a single object is seen against an otherwise uniform field (e.g. an
illuminated aperture with a dark background), convergence is virtually the
only clue to its distance that is available, if we ignore the extremely feeble
?(Frorn Ilelmholtz (I) Physiologis:he Optik. 2nd Ed. Leopold Voss, Hamburg and Leipzig, 1896.)
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approvor Release 2003/05/14: CIA-RDP78BW171A000600070052-0.
A 6Q AQ (b) RelativePositions of images on two retinae
showing disparity BLBR.
effect of accommodation, and if there are no subsidiary monocular clues to
help, such as a knowledge of the size of the object. Even convergence is a
very unreliable factor for the accurate determination of the distance of an
isolated object, perhaps because of rapid adaptation to the tension in the
muscles, but the simple experiment of observing the apparent change in size
of an object when a small-angle prism is held base outwards in front of one
eye, so requiring increased convergence for fusion, demonstrates with certainty
that convergence is a factor in space perception.
The smallest distance by which one object can be seen to lie ahead of
another depends very much on the observing conditions and the nature of the
two objects. The highest stereo-acuity will only be attained when the two
objects are well illuminated, are sharply defined and clearly contrasted against
the background, and are seen close to each other. This last condition ensures
that the retinal disparity can be registered with maximum precision using the
foveal area of the retina. Expressed in terms of angular disparity, a stereo-
acuity of 2 seconds of arc has been recorded under optimal conditions, but if
a value of 10 seconds of arc is taken as typical of normal viewing conditions,
this implies that for objects 5 ft. from the observer, there are 15 discriminable
planes per inch, and 4 discriminable planes per inch at a distance of 10 ft.
away. For two objects farther away than about 1,200 yards, it is impossible
on the basis of binocular parallax alone to distinguish which is the nearer,
however far apart they may be.
Good stereo-acuity may not necessarily be associated with an ability to
make accurate estimations of the finite difference of distance between two
objects. This calls for a different type of visual dexterity and a different type
of training, although no doubt under conditions where the acuity would be
poor, the estimation of differences would be reduced also.
Monocular Clues to Distance
It is quite possible to design experiments in which an observer using one
eye only virtually loses all ability to discriminate distances. Such experiments,
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approva&f or Release 2003/05/14: CIA-RDP78B 171A000600070052-0
however, are not typical of normal observing conditions, in which monocular
clues may contribute quite vitally to depth perception (Gibson, 1950); these
clues cannot be ignored in photogrammetry.
There is angular size-the nearer an object, the larger its angular subtense.
There is perspective, a factor-which is most apparent where some relatively
simple geometrical pattern exists, such as the converging of parallel lines or
the diminishing pattern of a tessellated pavement. Allied to this is the texture
gradient of a surface, in which the texture becomes vanishingly fine as the sur-
face recedes into the distance, whether the texture is made up of chips of a road
surface, the lumps of earth in a ploughed field, or the heads of the people in a
crowd.
Motion parallax is one of the most important monocular factors bringing
out the solidity and three-dimensional character of a real scene as the observer
moves his head, and its absence is perhaps the most significant difference
between a stereo photograph and a real scene. Other clues are the overlapping
of one object by another, the higher elevation of the more distant objects, the
information provided by the lighting and shadows, the aerial haze overlaying
the distant landscape, the sharply defined, clearly contrasted, near object seen
against the hazier distant background and so on.
In a normal scene, the monocular and binocular clue's are nearly all present
in some degree, and as they are usually in harmony, they combine to produce
a most compelling sense of depth and three dimensions. Considered indi-
vidually, binocular disparity is the primary and pre-eminent contribution, but
each factor plays its part. Under special, usually restricted, viewing conditions,
any one of them may become the dominant factor, while in trick situations,
some of the clues may apparently be in opposition to others. In such circum-
stances, the brain may require a considerable time to find the solution to the
visual problem presented to it.
PART It. APPLICATIONS TO PHOTOGRAMMETRY
General Viewing of Stereophotographs
When large-area stereo prints are being viewed, such as the prints prepared
by the anaglyph method in red and blue-green and viewed through corres-
pondingly coloured filters, or by the vectograph method and seen through
polaroid viewers, it is interesting to consider the relation of the horopter to the
judgment of the relative heights of widely separated objects in the photograph.
When a given point is being fixated, the apparent frontal plane, as already
mentioned, may not be a plane normal to the direction of view, and may
indeed by a curved surface either convex or concave to the observer, depending
on the viewing conditions. This would mean that objects having the same
disparity in the photograph would not necessarily appear to be in the same plane.
In an aerial photograph of, say, hills and valleys with no auxiliary features
which might help to define the horizontal plane, the curvature of the horopter
could be a potential source of error in judging any gross curvature or contour
changes of the surfaces.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approve46For Release 2003/05/14: CIA-RDP78Bi16171A000600070052-0
Of course, stereo pictures are not viewed by steady fixation on just one
point, but are scanned over their whole area. In that case, no single horopter
surface is involved and convergence clues help to interpret equalities and
differences of disparities. Most of all, however, the photograph has an edge
which in general is a rectangular edge or frame, and perhaps this more than
anything else helps to provide a reference plane relative to which the height
and depth of the hills and valleys can be compared. The rectangular shape
of the frame is not a specifically binocular factor, and we have here an example
of a monocular clue-shape or form-playing an essential role in a stereoscopic
problem.
Since the viewing of photographic prints under the conditions described
here is mainly carried out for qualitative rather than quantitative studies of
contours, the problem that has been raised is not of major importance in this
context. It is nevertheless of interest and is a useful introduction to an
analogous problem in the viewing of stereo-pictures in contour-plotting
machines to which reference will be made later.
LEFT
EYE
RIGHT
VIEW
RIGHT
EYE
LEFT
VIEW
Fig. X. Diagram to illustrate principle of viewing stereo
photographs with views transposed.
One further small point may be made in connection with the reproduction
of stereo-pairs side by side for viewing in a hand stereoscope. These are
normally arranged with the right-eye view on the right and the left-eye view
on the left, thus being ready for viewing in a stereoscope. Some people can
successfully relax their convergence while maintaining their accommodation
and are thus able to fuse and view these stereo-pairs without the aid of any
auxiliary optical system. This is a useful attribute which many more people
would like to be able to share and could almost certainly do so if the photo-
graphs were merely transposed, with the right picture on the left and vice
versa. With this arrangement it is only necessary to converge on to a point
between the card and the observer, as shown in Fig. X, for the pictures to be
correctly fused and seen in relief. This enhanced convergence is relatively
easy to produce, especially if the photographs are at first held well away from
the observer and he interposes his finger mid-way between himself and the
prints, to assist him in attaining adequate convergence. Fig. Xl provides a
transposed stereo pair on which the reader can make the experiment for
himself.
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070052-0
 Approver Release 2003/05/14: CIA-RDP78BQ A71A000600070052-0
Fig. XI. Transposed stereo photographs for viewing by the principle illustrated in Fig. X.
Binocular Viewing Systems
Many observers find great difficulty in making the two fields overlap
and in securing fusion when they look into a binocular instrument such as a
binocular microscope. The same is true when people look into the contour-
plotting machines used in photogrammetry, although since they are employed
on a much more restricted scale, fewer reports of such difficulty are encountered.
If the axes of the viewing systems are converging, fusion is more readily secured,
and it is evident that the cause of the trouble is an inability to relax the
accommodation and the convergence sufficiently for the axes of the eyes to
be parallel, the condition required for fusion if the eyepiece axes are parallel.
It is unlikely that there is a simple, or even single, reason for this difficulty,
which is apparently associated in some cases with a virtual paralysis of
accommodation (Inst. Geo. Nat., 1942). In the case of the microscope, the
fact that the observer knows that the object on the microscope stage is close
to him gives rise to a natural tendency to converge; indeed, where a binocular
microscope is being used as an aid to micro-manipulation, it is necessary that
the enlarged,image should appear to be located at a point in space near to the
operator, otherwise there will be confusion between the sense of touch and
the sense of sight. Perhaps the proximity of the eyepieces tends to induce
accommodation, and E. F. Fincham has some evidence that the smallness of
the field of view may enhance the effect. With instruments such as binoculars,
which are used out of doors and on distant objects, fusion difficulties are
generally much less pronounced.
Whether provision should be made for the axes of contour-plotting
machines to converge is a matter for argument. A period of training in the
use of these instruments is always necessary, and one of the aims of the
training is to develop greater freedom and flexibility of the accommodation
and convergence. After such training, fusion becomes easy, and it can be
argued that, since relaxed accommodation and convergence are the generally
acknowledged conditions for minimum visual fatigue, parallel optical axes
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved'1*r Release 2003/05/14: CIA-RDP78BO 71A000600070052-0
of the eyepiece system are correct for prolonged use of the equipment. On
the other hand, since this tendency towards convergence does exist with some
observers when first using the apparatus, it may be that the most restful
condition for them would be at least a slightly converging system.
Since the contour observations involve only relative settings between the
target and the aerial photograph, the convergence of the axes would not affect
the actual contour setting, although it might affect its precision. On balance,
the best arrangement would seem to be either a slight permanent convergence
of the axes, or some degree of adjustment to the convergence to suit the
individual user. Under no circumstances should the axes be set with any
permanent divergence, as this can lead to definite eyestrain. If a permanent
convergence is provided the angle between the axes should probably not
exceed 2 ?.
No question arises about parallelism of the axes of the eyepieces in the
vertical plane, since the eyes have only a limited independence of movement
in this plane. A deviation of 2 ? is the most that can be permitted and a
more accurate alignment than this is desirable, except that some adjustment
may be required to compensate for abnormalities of vertical muscle balance.
This tolerance refers to the axes of the viewing system as a whole. The
photographs themselves will have to be mounted more accurately than this
relative to the targets, although even here advantage can be taken of an effect
sometimes described erroneously but vividly as `retinal slip.' The fusion
process in the brain is evidently sufficiently fluid to overcome slight differences
of alignment up to a few minutes of arc between 'different points in the two
views, so that, provided there is sufficient fusion stimulus present in the form
of obvious identity of the relevant parts of the two photographs, both target
and photograph will be seen stereoscopically in spite of some differences of
registration.
In addition to errors of registration in the vertical and horizontal planes,
there may be an error of rotation about the axis of the system between the
two pictures. Fusion can usually be maintained in spite of a rotation of
even a few degrees, but this will depend on the subject of the photograph
and will in any case give rise to considerable eye-strain. A rotation of more
than 2 ? would be excessive, and a smaller tolerance than this is desirable.
One or two other points in the design of the viewing system deserve
attention. Eyepieces of optical instruments are usually designed so that the
exit-pupil is only a short distance (perhaps 4 or 2 inch) beyond the last lens
of the system. This means that a person with normal vision can bring his
eye close to the eyepiece with his eyebrows lightly touching the eye-ring and
with the exit-pupil of the eyepiece more or less in the same plane as his eye-
pupil. Under these conditions he will see the complete field fully illuminated
without any vignetting.
Persons who have to wear spectacles, however, are at a considerable
disadvantage, since they cannot bring their eyes sufficiently close to the eye-
piece to secure coincidence of exit-pupil and eye-pupil. As a consequence,
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved'r Release 2003/05/14: CIA-RDP78BOUflA000600070052-0
the field of view is restricted and its boundary is diffuse instead of being seen
sharply defined by the field stop. At least one firm, Messrs. Ross Ltd., has
recently taken steps to meet this situation in the case of prism binoculars,
since in their Spectaross binoculars special eyepieces have been designed
incorporating a large-diameter eye lens, an exit-pupil located farther from
the eye-lens, and adjustable eye-rings and a forehead rest to enable the bino-
culars to be used in comfort and with a complete field of view whether the
user wears spectacles or not. This principle might well be extended to other
optical instruments, and possibly to contour-plotting instruments, although
where a person requiring a spectacle correction is to use a particular plotting
machine over an extended period of time, a better solution would no doubt
be to incorporate his correction in a lehs attachment to the eyepiece.
Another fitting which is used in certain instruments designed for visual
research is a mouthpiece mounting on to which the observer bites and by
means of which his eyes can be centred and fixed relative to the exit-pupils
of the instrument. An example of such an instrument is shown in Fig. XII
(Wright, 1946). This device is especially desirable in photometric and colori-
metric observations, since the visual efficiency of a bean of light'is very much
reduced if it is off-centre as it passes through the pupil of the eye and, there-
fore, is not incident normally on the retina. The image is also subject to
Fig. X11. A binocular-matching photometer incorporating a mouthpiece mounting to locale
the observer relative to the exit-pupilr of the instrunrrrrt.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved Release 2003/05/14: CIA-RDP78B05 iA000600070052-0
some unsymmetrical aberrations which might possibly affect the settings made
in photogrammetry. A dental impression is less unpleasant to bite on than
it sounds, and it might be worth while to test its merits in contour-plotting
observations.
The functions of the field stop in a binocular eyepiece also deserve
attention. The field stop and the image being viewed are normally in the
same conjugate plane and both will be seen in sharp focus together. In
theory the binocular images of the stop will be seen fused and will occupy
a definite position in space, but the possible curvature of the horopter illus-
trated in Fig. VII suggests that the field stop may not appear in the same
frontal plane as the target.
Now, in the case of stereo-prints, it has already been suggested that the
bounding edge is important in providing a plane of reference for judging
the contours, and the same may be true in an eyepiece system. On the other
hand, it might prove a distraction and affect the setting of the target relative
to the photograph. Again, when heights are being determined by changing
the horizontal separation of the photographs, the observer has the convenient
impression that it is the target that is approaching or receding, but if the
target were a rather elaborate structure-if, for example, the field stop and
the target were obviously associated together-the scene itself might appear
to approach or recede. This would hardly be such a mentally satisfying
conception or so conducive to accurate or rapid observations.
At least further thought might well be given to the field stop, and the
advantages and disadvantages of a diffuse boundary to the field might be
considered. Perhaps, also, tests with an illuminated surround might be worth
while.
The Visual Task in Contour Plotting
The speed at which stereo-plotting machines can be operated is a tribute
to the ability of the observer's eye and mind to assimilate complex information
and emerge with a simple interpretation. This ability is acquired by training
and experience and it is of great interest to analyse the nature of the visual
task which the operator has to perform, provided the operator himself does
not become too self-conscious of his visual processes. An analysis may
have a direct bearing on the design of the apparatus and on the training
programme, but if the observer were to think too much of what he was doing
and how he was doing it, the results might be disastrous to his success as a
contour plotter.
An adequate analysis of the task almost certainly requires a knowledge
of the eye movements of the operator, but we can only infer these, since no
direct records have been obtained during the act of contour plotting. We
can assume that for a considerable fraction of the time, fixation is maintained
on the stationary target yet, as shown in Fig. IV, this does not mean that the
eye is absolutely steady; random movements of a few minutes of arc are certain
to occur. In addition, there will be more conscious excursions to various
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070052-0
 Approv or-Release-2003/05/14 CIA-R-E3P78E 1.71AO0060007005 Q
nearby points of the stereo photographs in order to obtain as accurate an
idea as possible of the three-dimensional structure of the scene in the immediate
locality where the contour is being plotted. No doubt there will also be
occasional excursions of the eyes to the more marginal parts of the field to
secure a general idea of the contour trends. What is probably not known
at all at the moment, is how far the sweeping of the photographs across the
stationary fixation target may induce some degree of the `pursuit' type of
eye movement. If the target on which fixation was being maintained were
the moving object then undoubtedly there would be some pursuit movements,
although this would depend on the speed of traverse of the target.
Another question that arises is how far hunting occurs in contour plotting.
The operator has to look slightly ahead of the point actually being plotted
at any given moment and has to judge in which direction across the terrain
A he height is remaining the same. He cannot make the most accurate judgment
of this until the target is alongside the new point and the chances are that
there will be a slight error, which he can then correct. This will be repeated
for the next point and so on, so that the contour line as plotted must consist
of minute deviations from the true contour. It is a fascinating problem in
what might be termed `mental mechanics' to reduce this hunting to a minimum
and to design the controls and the plotting mechanism so that their inertia
matches the mental characteristics of the operator. One of the functions of
the training of the operator must be to give him adequate confidence to plot
the contours smoothly with the minimum of hunting and to adapt himself
to match the equipment he is using.
One potential source of error, and a likely cause of differences of setting
between one observer and another, is the influence which monocular clues,
such as the effect of light and shade, the sharpness and contrast of the outline,
the interpretation of perspective and so on may have on the judgment of the
contour. The photographic quality of the picture, the optical magnification
relative to the picture definition, the density of the photograph relative to the
blackness of the target, are also attributes of the system that may affect the
purely stereoscopic determination of the contour. For example, if the mag-
nification is too great, so that the graininess of the emulsion is revealed, a
marked flattening of the scene will be produced because the texture gradient
of the grain will be zero; empty magnification may also make the picture more
difficult to interpret. -
Fatigue or poor health of the operator is known to influence the setting
he makes, and the tendency is for him to dig the target into the ground when
he is tired. Reduced precision of setting might well be expected under these
conditions, but a consistent error in a given direction is less easy to explain.
The most likely explanation appears to be that the binocular sense becomes
less dominant, and monocular clues carry more weight; in particular, the
sharpness, blackness and high contrast of the target may give the impression
that it is nearer to the observer relative to the terrain than its disparity would
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070052-0
 ApprovadrFor Release 2003/05/14: CIA-RDP78MO 171A000600070052-0
imply. To offset this, the stereoscopic setting would then have to be in the
opposite direction, that is, the target would be `dug in' to the ground. More-
over, if each observer gives his own characteristic weighting to the monocular
cities, this would account for any consistent difference or personal equation
that might be found between one observer and another.
Auu.viliar y Observations in Photogranunctr)'
In addition to the actual observations made with photogrammetric
equipment, binocular vision is, of course, used in the actual drawing of the
maps, both in following the drawing point attached to the machines and in
elaborating the maps on the drawing board. Close fine work of this quality
can lead to eyestrain, and it may be worth while to refer to the use of prismatic
spectacles that are recommended in certain industries where prolonged
accommodation and convergence on near work are involved (Weston, 1949).
The principle is simply to provide prism and lens power such that, for a given
working distance, both the accommodation and convergence is relaxed as if
the observer were viewing an object at infinity. These spectacles may be of
quite high power for viewing objects only a few inches away, in which case
they correspond somewhat to a binocular watchmaker's glass or they may be
designed for working at ' metre or 1 metre away. At least one manufacturer
provides-them in a convenient form made with the frames and lenses in plastic.
. The drawing and study of contours lying very close together also involves
the limited precision with which the visual axis can be defined, as discussed
in Part I. If a row of dots or series of lines are placed so close together that
they have an angular separation at the eye of the observer of less than about
3 or 4 minutes of arc, it is found that the observer can no longer maintain
his fixation on any one of them, or identify or follow a particular line. If
map drawing should involve this precision of working then eyestrain is almost
certain to ensue and steps should be taken to provide optical magnification
to relieve the task.
Chromatic Aberration and the Multiplex Process
In the Multiplex process the stereo pair of pictures are projected through
red and blue-green filters, and the operator wears viewers with similar red
and blue-green filters over the right and left eyes respectively. With the
correct spectral transmission of the filters this arrangement ensures that each
eye sees only one picture.
The fusion of differently coloured views to give three-dimensional relief
is quite effective, although some persons find difficulty and even serious dis-
comfort. A factor, however, which is often ignored is the chromatic aber-
ration of the eye. The optical system of the eye is quite uncorrected so far
as chromatic aberration is concerned and its decrease in power (expressed
in dioptres) with increase in wavelength is illustrated in Fig. XIII. (Ivanoff,
1953.) Typical spectral transmission curves of the filters used in the Multi-
plex system. are shown in Fig. XIV (Pestrecov, 1953) and the wavelengths
Approved For Release 2003/05/14: CIA-RDP78B05171A000600070052-0
 : Approved-Release 2OG3/05/1,41 CIA-RDP78B0> 1-A00060O00-7.0052-0-
WAVELENGTH (mjt) WAVELENGTH (m?)
Fig. X/!!. Chromatic aberration of the Fig. XIV. Special transmission curves of typical
eye as shown by its decrease in power red and blue-green filters used in the Multiplex
with increase in wavelength system
(after Ivanofr)
0.48w and 0.621, can be taken as representative of the hues and average wave-
length transmissions of the two filters. Fig. XIII then shows that there is
a residual difference of power of about 0.75 dioptres between these wavelengths.
If this difference is not corrected, either by incorporating a negative lens
with the blue filter or, as is to be preferred, a positive lens with the red filter,
then two results will follow. In the first place, it will mean the exercise of
differing amounts of accommodation in the two eyes if both views are to be
seen sharply in focus; this is likely to be a potential source of eyestrain and
of reduced precision of observation. Secondly, for persons about 45 years
of age and over a significant loss of accommodation begins to be experienced
and for many such people there will not be sufficient power for a sharp red
image to be focused. This would prevent any accurate observations and
even any effective three-dimensional reproduction of the scene.
It would be of interest to know whether tests have been made to compare
colour defective observers and persons with normal colour vision in their
ability to use the Multiplex equipment. Colour defectives will still, of course,
be affected by the differing refractive power for the short and long-wave-
length light, but they will be less troubled by colour differences when trying
to fuse the two pictures. Probably the class of colour defective known as
deuteranope (Physical Society, 1946) would prove most efficient at the
observations, since they have poor red-green colour discrimination yet their
light sensitivity to different wavelengths is similar to that of normal observers,
hence the two pictures would have the same lightness as the normal.
Colour Stereoscopy
Finally, an effect should be mentioned which can give rise to an apparent
difference in depth for differently coloured objects lying in the same plane.
This phenomenon is due to a decentration of the eye pupils in conjunction
with the chromatic aberration of the eye. (Duke Elder, 1932, p. 1067.) Fig.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved Forrelease 2003/05/14: CIA-RDP788051-4.kO00600070052-0
XV illustrates the paths of red
outwards, leading to it lateral (land
isp lbccmntsof for tilee
the retina reds and displacement is i
. This n a direction corresponding pupils decentred
for the red object located in front of d blue images displaced inwards, the order will be reversed blue to the disparity tile , but for persons with pupils
TYg? X : The sepuruNorr ill denrh ?r_ .. '
Rrn
c uocrrati?rt uf!. luPupilcle
This explains why roads marked the eye cerrlrptiort may appear to be slight in hig
colours O11 a map
ly about or below the have some effect on the correctness of b e elt' surf rftcice of rated tile ill,
wise is unlike( to Multiplex observationp It might
come to be Y be of much practical importance unless colour photographs in aerial surveying. ~ but otlter-
Pupils of the used
In that case, the centrin otographs
assume much greater plotting machine relative to the eye g of the exit-
h renter importance and pupils of the observer
piece mounting to fix the observer's head might most ssent aaddition might
of a mouth-
Research on stereoscopic vi ACKNOWLED NTS
Im erial GMn
assistancear liege and is supported b Progress in the Technical Q
discussions I grateful have hl acknowledgedY I should from the Medical Researc ptics Section
E. H. also like h Council, of the
Thompson and Dr, during the last few months with a brof Howled This
of the Institute of O E. A. Miskin of gc the very helpful
Mr. J. Cruset of of Pht l o to y' Mr. P. University Coll c e, People including Professor Insti members of the stair of the Henr G. Mott of ar A.j, S, ~Mr. E. F. nd m
tut Switzerland, It has onl b Phique National , Paris, and Mnd b Acrosurve
Y Wild Survc in ? M? A, J. Co. Ltheini Is tree
oth othcr
better understandin y been through these discuss onsnthats l Supply of the visual problems in pply d., o Heerbrugg,
photogramnictry have been Lteen able to obtain a
(I) Amcs, A., Oglc, K. REFER
(2) Aes,Elder, N. and Gliddon, LNCES
(3) Gibson, W. S?? 1932. Tex!-book of o 1u193! rol1 oopt Soc. , 538
Boston). J. J'' 1950. Tire Perception o p Irrr? 22
f,t~he IYisrta/ Wordl? 1. (KimptonaLondS
(4) /o, Institut Gcographique National 1c on).
t e! d'erurainpnut, X42. (Houghton Mifflin
(5) iou clvano(1 Notice Provisuire sus /e Stereoscope de dernouCra-
A?, 1953.
(6) Lord M... 953 Les Aber,01lous rle !'Dei! (R1948. (7) Lord M. P. ,and W Proc. Phys? Soc. 61 439 ev, d'Qptique: Paris ,
Pit's. Soc. Re right, W, D,, 1948, )
p Prog, Plrys., 13, 1. Na,,,,?e, 162, 25. 1949
(8) Lord PCsUCCOV K
., 1953. grtr J. O , Nntrrre, 163, 803.
(9) Physical Society, 1946. Re Plwfe ti30 e 366. ]950,
(10) Walls, G. L., 1952. p? ort Defective Colour l133, t itt lnrlustrv.
(12) 11
WestonI. C., 1949. ASi /t Optom. Monog,. No. 133.
Wright, , 1 W, D , 1946. R sea chest on No~ n~uaru. (ewis: )
London), 1951, Proc. Pit s. London
y Soc. B., 64, 289 fective Colour Yision.
(Kimpton:
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approv or Release 2003705/14-. CIA-- RDP78 `17'1A0G060007M2--6 -
DISCUSSION ON PAPER BY PROFESSOR WRIGHT
Dr. Miskin said he was most interested in the point regarding definition and
made the statement that tiredness or lack of definition of the photograph would
lead to a `digging in' of the-actual measuring mark. He asked whether Professor
Wright had considered this fact in the relationship of the photographs themselves.
If the photographs had soft spots, as happened with some present-day photographic
equipment, would 'digging in' be likely to occur where there were soft spots and no
'digging' where the definition was good?.
He was also interested in the convergence factors which had so adequately been
put forward and he thought the point was well-illustrated. He remarked that when
Professor Wright had visited him he noticed that the Professor used a prism to
obtain some degree of convergence when viewing through the instrument.
Another important point was in connection with the anaglyphic method of
viewing as he believed there was some delay in reception in the brain of red and
green light, and he wondered what effect that slight time lag might have in the
perception of the three-dimensional model.
The interesting part, so far as the photogrammetrist was concerned, was the
ability to stop fatigue, and he asked Professor Wright if he had any actual figures
for fatigue values comparing monochromatic lighting and anaglyph lighting.
Professor Wright answering the first point of Dr. Miskin's question said he
thought that with soft spots there might be a tendency to 'dig in'. He understood
the question of photographic quality of the pictures might be made a subject for
international study and standardisation.
Regarding the time lag with red and green pictures, Professor Wright said that
this could cause the pictures to appear to jump about, although he had never
observed the effect. There was also a certain amount of rivalry in the two colours
which caused difficulty in fusing.
He said he had no figures regarding fatigue, because he was not really working
close enough to the problem, but fatigue was a difficult thing to measure.
Mr. Fish asked Professor Wright whether he had any information regarding
the phenomena which some people found when viewing stereoscopically, that the
resultant picture had better definition than either of the two single pictures. He
wondered whether it was because the pictures, at the scale viewed, tended to show
the grain. It was a fact that quite a lot of the area of the picture looked at must
have differences in definition between the two eyes, because the 40 per cent overlap
was in the outer part of one and the centre of the other photograph.
He said that use had been made on occasions of stereoscopic photographs
taken from forward looking obliques, which depended on the forward motion of
the aircraft for stereoscopic effect, and these had produced some queer results.
He pointed out that in many instances there must be differences in scale between
two pictures, but the eye still appeared to give stereoscopic vision and he wondered
whether this had ever been investigated. Another thing arose under the same
conditions, on occasions the angle of convergence was very great and Mr. Fish asked
Professor Wright whether he could give any idea what the limiting angle was for a
pair of eyes to see stereoscopically.
Regarding the photographs of the saccadic jumps in the movement of the eyes,
he asked whether there was any relationship between the two eyes or if they were
random between the two.
Professor Wright agreed that better definition was obtained with two photo-
graphs, and must in part be due to the averaging out of the grain effect.
Professor Thompson asked whether there was a similar improvement when a
natural object was examined with two eyes.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approved Release 2003/05/14: CIA-RDP78BOSOr 1A000600070052-0
Professor Wright said our visual acuity when looking at perfectly defined objects
was better with two eyes than with one, no doubt because the limitations of the
retinal mosaic and retinal response were also to sonic extent averaged out.
Mr. Fish remarked that it did appear when looking at two poor pictures you
got one better one out of them.
Professor Wright referred to his slide showing dot patterns and said his point
in showing that was to emphasise that in photography where there was grain, if there
was too much magnification so that the grain became apparent, it tended to produce
a flattening effect of the picture.
Regarding the difference in scale, Professor Wright said that similar trouble
was experience([ with people who had one eye giving a different size image on the
retina from the other. This subject had been studied in America and he knew of
experiments by Lord Charnwood in this country.
The limit of convergence depended on age and focusing power. It was possible
to get a three-dimensional effect where the subject was converging to a distance of
10 inches, or nearer.
Mr. Fish said he was thinking in terms of the fact that the object was being
viewed from widely divergent positions.
Professor Wright said it was a question of how much disparity one could stand
and still fuse.
Regarding the saccadic movements, the eyes did go in harmony to within close
limits. The recent work carried out by Dr. Mary Lord on binocular movement
records confirmed this. He said that with photographs taken from widely separated
points, the views might be so different that fusion was very difficult or impossible,
but as this would depend on the scene being photographed, it would not be easy
to give a value to the limiting angle. Angular disparities of several degrees could
be overcome by change of convergence, but if the disparity varied too drastically
from point to point of the photographs, the three-dimensional effect that could be
observed would be very poor.
Mr. Dawe said he had one or two things to comment on. Firstly, the question
of whether an operator was actually 'hunting' when searching for contours. He
thought he was, but the more practised and experienced the operator the less
`hunting' was done. The particular set-up in the instrument (that is the positions
of the pictures relative to the two separate reference marks in the binocular instru-
ments) indicated the correct plane. A quick check could often be made by closing
each eye alternately very quickly. If the mark appeared to jump it indicated that
the mark was in the wrong plane whereas when the correct position was occupied
there was no apparent movement. The actual process of plotting the contours
must inevitably entail a certain amount of `hunting' and feeling when the dot was
about to split into the ground or come 'off the deck.'
No planned study of fatigue had been made, but operators on anaglyph work
asked for a change more often than those on the binocular instruments. If necessary
convergence down the telescope of an A5 or A6 could be engineered to a certain
extent on the present models by the small adjustment of one of the prisms without
affecting the internal set-up of the optics.
Professor Thompson referred to the viewing of repeated patterns by over-
convergence of the eyes. The fused patterns then appeared considerably closer to
the observer than the original object. Although it was said that this was due to
the convergence it was unquestionably due partly to the presence of other objects
in the field of view which, although they appeared double, gave a reference plane
in depth.
Professor Wright said he had given demonstrations recently to show that some
appearance of depth could be observed even when an object was seen double, but
this had, of course, been known for a long time.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approver Release 2003/05/14: CIA-RDP78BQiM71A000600070052-0,
Mr. Cook referred to the request for convergence in the binocular machine
and asked if Professor Wright thought it was more of a mental request than optical
necessity for the eyes to converge on a point hequired when looking into the binocular
attachments.
Professor Wright said it was difficult to separate the two, the eye and the mind.
It was difficult to dissociate what you saw and associated with past experience and
maybe what your nuiscles and lenses were demanding. He would not like to
distinguish between the two.
Mr. Richards said that regarding convergence it appeared that the eyes were
more content or rested if they were fixed on some distant point. In a binocular
instrument was it not an advantage if such instruments were non-converging so that
no strain or undue work was given to the eyes of an operator when the sight was
kept near parallel.
Professor Wright agreed that this was a reasonable argument, but if the act of
looking through a binocular eyepiece produced a natural tendency to converge, the
best solution might be to have a very slight convergence. It might at least be helpful
to sonic people and for most beginners if the convergence could be adjustable.
Mr. Richards wondered whether this would have any bearing on the fact that
after an operator had worked for a while he tended to `dig' on the model because
the eyes wanted to converge but they were being forced apart to look at the floating
mark and in doing so, when looking at the model later on in the day, it would appear
that the reading was lower on that point, and it would seem that the eyes were being
forced outwards.
Professor Wright said he had been trying to think out an explanation for this.
`digging in' but he did not see how it was going to work on that explanation.
Whatever applied to the dot would apply to the photograph, so there would be no
residual disparity. His suggestion of the effect of the sharpness of the dot relative
to the less sharp photograph seemed to provide a more plausible explanation, but
he had no proof of it.
Mr. Attwcll mentioned the difficulty initially in fusing the floating mark due to
the indefinite nature of the mark. Disregarding the possible objections that the
experienced operator might have, he asked whether Professor Wright thought if
the floating mark was more definite than a dot, such as a split cross, fusing in the
initial stages of training would be easier.
Professor Wright thought this was very likely and said that if there was a bit
more structure to the object one was trying to fuse it would be done mor(, readily.
He imagined that in the end the dot virtually acquired a personality and was not
just a dot, but something one looked forward to looking at, but until that stage was
reached a more complicated pattern would provide a stronger stimulus to fusion.
Colonel Gardiner asked about the effects of colour. With coloured photo-
graphs would the marks 'dig in' more in sonic portions of the photograph than others?
Was colour photography going to be a nuisance and black and white the best thing?
Professor Wright said he did not know and he had not yet carried out any
experiments with colour photographs. He thought there would be a gain from less
graininess in the picture with coloured photographs. Colour would also probably
help because it would add to the reality of the situation.
Colonel Gardiner asked what would be the effects of the different colours.
Professor Wright said there would not be the anaglyph trouble of the two eyes seeing
different colours. Errors due to the `colour stereoscopy' effect described in the
paper might, however, be important unless the exit pupils of the instrument were
accurately located on the visual axes of the observer's eyes. This might make the
use of a mouthpiece mounting essential.
Professor Thompson said that more work could be done on the tolerances
permissible in the direction perpendicular to the eye base. For purposes of corre-
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0
 Approvq&For Release 2003/05/14: CIA-RDP78195171A000600070052-0
spondence setting it was not sufficient, he thought, that the photographs could merely
be fused.
Professor Wright agreed that the tolerances he had suggested for fusion of the
pictures would be much too large as tolerances for the location of the photographs
in a plotting machine.
Professor Thompson thought it had been a most interesting paper. Photo-
granimetrists were the principal users of stereoscopy in a serious way and there was
clearly a need to enlist the support of men such as Professor Wright who had studied
the physiology of the subject. lie hoped that this paper would lead to further
collaboration. Professor Thompson concluded by thanking Professor Wright on
behalf of the members of the Photogrammetric Society for having taken so much
trouble in preparing and delivering such an interesting paper.
Approved For Release 2003/05/14: CIA-RDP78BO5171A000600070052-0