(Sanitized)PROPOSAL

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78B04770A000200010049-7
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 15
• Document Creation Date: December 28, 2016
• Document Release Date: July 1, 2004
• Sequence Number: 49
• Publication Date: January 1, 1966
• Content Type: REQ

Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
C, Vj 0 -ex-CA
A_
C~_h S W `~ S W e h e e l v h
-t-k c 9 O s ~,~,~ !~ S S \1vQ Sko~ t
Declass Review by NGA.
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 % Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
t
`e h S ~ t S ~ V( ON. , , -P..._// G.. .
-1--v- ---,O D --
W\
- ~-, cam- ?-- . ~. _
~;
- ~..L?_ r ~ ,...t.~,. ?.~ V~. ~.~:..0 C,
451-1
Approved For Release 2004/07/29 : CIA-RDP78B04770'A000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
17
.~. it
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
PROPOSED CHANGE OF SCOPE
For
MULTIPLE IMAGE INTEGRATION STUDY
August 1966
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
MULTIPLE IMAGE INTEGRATION STUDY
?
1. Introduction
The first phase of this study was concerned with an analysis
of the problems of image integration, and the initial design for a
three-input image integration viewer/printer that would enable a
practical evaluation of the benefits to be obtained from the combin-
ation of several images of the same object.
The hardware approach was based on the use of electronic
imaging and video processing, using the ARES image correlation
techniques. Relative geometrical distortion between the inputs was
to be cleared by changing the shape of the cathode ray tube scanning
rasters. Because of the requirement for high resolution, a slow scan
print-out onto film was to be used; direct viewing of the integrated
image at high resolution was not possible. The resolution capability
of the proposed image integration viewer at 12X magnification was
estimated to be similar to that of a high quality photographic enlarger
with a limiting resolution in the region of 80 - 100 lines/mm.
For the purpose of image integration, an electronic imaging
system has the advantages that (1) high-order image distortions, even
those due to stereo relief, can ultimately be corrected, (2) a consid-
erable degree of video processing can be introduced. This is especially
necessary with material taken at different times of day or in different
seasons of the year, and may be desirable with multi-sensor material.
The main limitation of an electronic system is its restricted
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
bandwidth which limits the number of resolution elements that can be
displayed per second.
The only way of obtaining a high resolution direct viewing capa-
bility is by the use of an optical system. Since the interim report
on the Multiple Image Integration Viewer/Printer was issued (Dec.
1965) the feasibility of using an optical system to automatically
correct image distortion has been established. This method will
presently correct zero and first-order distortion, and offers direct
viewing of images with the ability to see detail at 150 - 200 lines/mm.
These advantages may outweigh the lack of video processing capability.
The problem of high order distortion due to different aspect angles
can be overcome very neatly in an optical system by taking the inputs
two at a time and using them as stereo pairs. All images would first
be normalized or reduced to the same scale, and then selected in pairs
as required by the photointerpreter. Direct superimposition of the
images is also possible, and experiments with various change detection
techniques could be made. A further advantage of an optical viewing
system is that color material can be used.
An automatic registration optical viewer is consequently a
very powerful tool with which to conduct an intensive study of multiple
image integration techniques.
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
?
2. EROS Tests
To provide some initial data on the feasibility of an optical
system for multiple image integration, some tests were conducted on
the EROS viewer. The material used was high-resolution, black and white
material comprising three coverages of the same area.
In order to see all detail on the input material, the maximum
EROS magnification of 24X had to be employed. At this magnification,
the precision of the X-Y film transport is not sufficient for critical
viewing. To provide optimum viewing, a further increase in magnifica-
tion to 50X or 10OX would be required. This would also improve
correlation efficiency, because at present much detail is beyond
the resolution of the correlation system.
Within the limitations imposed by the film transport and lack of
magnification, the results were very encouraging. Good correlation
was obtained between photographs taken at different times, and stereo
viewing of this material appears to be feasible. The conclusions to
be drawn from these initial tests are that
(1) An optical viewing system with image transformations is
suitable for image integration.
(2) A magnification of 50X to 10OX is required for integration
of high resolution photography.
(3) A highly precise film positioning system is necessary.
(4) An image correlator sensitive to high spatial frequencies
is desirable.
The EROS equipment in its present form would not be suitable for
this purpose without considerable modification.
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
?
?
3. Proposed Program
In order to evaluate the potential of multiple image integra-
tion, it is proposed that (1) a breadboard optical system to accept
up to three inputs be constructed, (2) an extensive study program
covering many aspects of multiple image integration be conducted.
The breadboard equipment would be intended solely to support the study
program, on which the major effort would be expended.
3.1 Breadboard Equipment
3.1.1 Requirements
In order to properly evaluate the various techniques which may
be applied to image integration, the following general equipment
requirements are established.
(a) Direct optical viewing, with a resolution of 150 lines/mm
or better.
(b) Automatic correction of parallax, scale, rotation and
anamorphic magnification.
(c) Superimposition of up to three images which have been
optically normalized.
(d) Stereo viewing of any two combinations of the three images.
In addition, the breadboard equipment should be flexible in
design so that the incorporation of optical or spatial filtering
could be investigated.
3.1.2 Proposed Equipment
In order to permit accurate and efficient image integration,
means must be provided to transform the images, optically, in order
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
0
to provide registration. Zero and first order corrections (parallax,
scale, rotation and anamorphic magnification) will be applied using
methods similar to those employed in the EROS equipment described
previously.
The error signals used to drive the optical and mechanical
elements will be derived by scanning each photograph with a vidicon
tube, correlating the resulting video signals, and analyzing the
relative distortion into its basic components.
The use of vidicons to produce video signals for correlation
considerably simplifies the optical design compared to the present
EROS system which uses a CRT scanner, and optical duplexing.
A block diagram of the breadboard equipment proposed is shown
in Fig. 1. The optical system is based on that of the EROS viewer,
with higher magnification capability of 25X to 10OX. The first
(objective) lens will have a magnification of 1OX, the zoom system
will be variable from .25X to 1X, and lOX eyepieces will normally
be used.
Photograph B can be considered the reference, with photographs
A and C capable of being correlated and transformed to match B. Rota-
tion and anamorphic correction are not required in the B leg, resulting
in considerable simplification of the optics. Individual light sources
are used for the photographs and neutral density filters may be placed
in each optical path to provide similar illumination levels.
Only csne
correlation system will be required, as each of the A
and C inputs are sequentially corrected to match B. Stereo viewing
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
of any selected pair, AB, AC or BC can be carried out or superposi-
tion of any two, or all three photographs can be viewed, by suitable
switching of mirrors.
Provision will be made for hard copy print-out of the super-
imposed image. This will consist of the image normally viewed in the
eyepiece, at a magnification of 2.5X to lOX relative to the input
film.
In order to simplify the breadboard equipment, it is proposed
that chips up to 2 x 2 inches be used with the film chip mounted on
a glass platen which is capable of being transported in X and Y by
a servo driven stage.
Previous experience with similar correlation equipment indicates
that registration of images to an accuracy of about 0.1% of the width
scanned can be achieved. Thus, in order not to degrade the 150
lines/mm resolution capability imposed, the area examined at maximum
magnification should be no more than 2 mm square for direct integra-
tion of imagery. Stereo viewing would allow somewhat more tolerance
on the field of view. A 4 - 1 zoom range should be adequate to
allow evaluation of image integration techniques.
Modulation transfer curves for the optical-integration viewer
are shown in Figure 2. Resolution measurements made on the EROS
viewer have shown that the resolution of the image transformation
system alone (zoom lens, rotation mirrors and two anamorphic lenses)
is about 60 lines/mm. With the objective lens of 2.5X presently
employed, overall limiting resolution has been measured at 160 lines/mm,
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 ?
0
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
-7-
which is approximately 2.5 x 60 lines/mm. This shows that the image
transformation system effectively limits overall resolution when low
power objectives of up to 2.5X are used.
With a high power objective of lOX magnification, the situation
is quite different. In this case, the limiting resolution of the
transformation system becomes 600 lines/mm referred to the input
film, which is far above the resolution of the objective lens. Under
this condition therefore, the overall resolution is determined largely
by the objective lens.
Figure 2 shows the MTF of a typical high quality enlarging lens
at 10X, together with the modulation curve of a typical aerial photo-
graph on 3404 film. The resulting MTF of the visual image, as seen
in the eyepieces is also shown. To determine the MTF of the hard
copy, the characteristics of the copy film must be taken into account;
a curve for type 8430 film is shown, together with the expected MTF
of the recorded image.
It can be seen that in order to see ground detail recorded on
the film at around 100 lines/mm, the optical system must have a limiting
resolution of over 200 lines/mm.
3.2 Study Program
The study program will constitute the major part of the pro-
posed work. The use of a high resolution optical system will enable
the effects of multiple image integration to be investigated for any
type or combination of input material. Because of the use of direct
optical viewing, the results obtained and any new techniques developed
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
0
will be directly applicable to a real photointerpretive operation
in which speed and accuracy are vital.
The main areas to be studied will be as follows:
A. Integration of two and three photographic inputs with appropriate
geometrical transformations into a single integrated output. Among
the factors whose influence on the quality of the final image must
be accounted for are:
(1) Aspect angles of the inputs
(2) Scale differences in original material
(3) Resolution of original material
The integration of black and white material with color photography
will also be investigated.
B. Integration of photographic inputs two at a time, using stereo
viewing. The factors to be investigated will be:
(1) What range of aspect angles can be accommodated by stereo
viewing.
(2) How accurately must scale and geometry be matched.
(3) What difference in resolution can be accommodated between the
two inputs.
(4) The effect of mixing black and white and color material.
C. Integration of multi-sensor inputs into a single output or a stereo
pair. Because of the difference in appearance of imagery produced
by different sensors, there is a considerable correlation problem
with this material. Because IR and radar systems do not produce
the same resolvable detail as an optical system, the additional
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
information obtained through multi-sensor integration is largely
in the form of cues which must be matched with recognizable objects
on a visual photograph. Methods of achieving this aim, by direct
superimposition or stereo viewing will be investigated. The poss-
ibility of color coding the multi-sensor inputs will be considered,
as will the use of flicker techniques.
The material required to conduct this study will consist of photo-
graphic coverage of several different areas obtained by several
missions at different times of the year.
Both panoramic and oblique material will be used with as wide a
range of scale and aspect angle as possible. Both color and black
and white coverage will be required. Multi coverage material comprising
photographic
of the same area will be necessary.
The imagery produced by active radar systems is primarily composed of
specular reflections and the detailed appearance of most objects varies
with the orientation of the line of flight. To properly evaluate the
integration of radar imagery with photographic imagery it is therefore
necessary to have samples obtained on several different runs.
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
 Approved For Release 2004/07/216 CIA-RDP78B04770A000200010049-7
O O Light
Source
Input Film
Objective
Lens
Zoom
i
Anamorph 1
Anamorph 2
0
Transformation
System
Stereo
Viewing
Scheme
Modification For
Superimposition
of Three Inputs
FIGURE 1 OPTICAL IMAGE INTEGRATOR
Approved For Release 2004/07/29 : CIA-RDP78BO47 OA000200010049-7
 Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
?
0
i1 ' i i'= L P Wro
e1,1 3404 tliM
IU7~x TOTAL
100 cycles. per mm
FIGURE 2 MODULATION TRANSFER FUNCTIONS
Approved For Release 2004/07/29 : CIA-RDP78BO477OA000200010049-7
oPT-c, L SYSTEM
Wi~"rl iox aa7~=Gr~
100