PROPOSED DEVELOPMENT TASKS FOLLOW-ON TO PROJECT(Sanitized)
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP78B04770A000200010002-8
Release Decision:
RIPPUB
Original Classification:
K
Document Page Count:
19
Document Creation Date:
December 28, 2016
Document Release Date:
July 1, 2004
Sequence Number:
2
Case Number:
Publication Date:
October 6, 1967
Content Type:
REQ
File:
Attachment | Size |
---|---|
CIA-RDP78B04770A000200010002-8.pdf | 763.58 KB |
Body:
STAT
Approved Foripease 2004/07/29 : CIA-RDP78B0477000200010002-8
PROPOSED DEVELOPMENT TASKS
FOLLOW-ON TO PROJECT
Proposal
Copy of Copies
6 OCTOBER 1967
Declass Review by NGA.
Approved For Release 2004/07/29 : CIA-RDP78B04770A000200010002-8
Approved Forelease 2004/07/29: CIA-RDP78B0477000200010002-8
PROPOSED DEVELOPMENT TASKS
SUMMARY:
In this proposal, eight tasks are described, all related to the development
of an Automated Stereo Scanner. These tasks are as follows, in our estimated
order of priority:
1. Extension of Stereo Scanner capability to a pointing camera
acquisition system.
2. Production of Correlator Test Photography.
3. Correction of higher order distortion in scanning system.
4. Servo Stability investigation.
5. Correlation with partially obscured images.
6. Search Simulation.
7. Determination of the optimum sequence for distortion removal.
8. Investigation of magnitude of incorrected second order Y-
parallaxes.
The first five tasks are of most importance in that the results will
significantly influence the performance of the automated stereo scanner. The
results of tasks 6 and 7 will also affect performance of the system, but to a
somewhat lesser degree.
The results of task 8 will determine whether a second-order optical
distortion correction capability is desirable for operator viewing comfort and
efficiency, and is therefore, somewhat forward looking in that the results will
affect the next generation of stereo viewing instruments.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For 110ase 2004/07/29 : CIA-RDP78B047701100200010002-8
%
TASK DESCRIPTIONS
AND
WORK STATEMENTS
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved Fortlease 2004/07/29 : CIA-RDP78B047.000200010002-8
?
TASK (1) - EXTENSION OF STEREO SCANNER CAPABILITY
The purpose of this task is to add to the capability of the stereo
scanner by developing the programs required for a pointing camera
acquisition system.
?
The stereo scanner control computer is currently being programmed
to handle panoramic and strip materials. An extension of its capabilities
to pointing frame materials requires development of a system model,
reformatting the INS data tapes and the development of specific control
computer programs.
A complete description of these tasks is contained in
9609-39, 15 May, 1967.
Document
WORK STATEMENT:
1. Development a model of the system from which distortion
coefficients can be calculated (General Purpose computer,.
Fortran Program).
2. Reformat the INS data tapes to allow their use in the above
program.
3. Development of control computer programs specific to the
pointing system.
Approved For Release 2004/07/29: CIA-RDP78604770A000200010002-8
Approved Foillease 2004/07/29: CIA-RDP78130471,00200010002-8
TASK (2) - CORRELATOR TEST PHOTOGRAPHY
The development and performance evaluation of the automated stereo
scanner will be considerably aided if a pattern or series of patterns were
available which would allow some determination of correlation reliability.
The use of such patterns is analogous to the use of resolution targets.to
evaluate the optical performance of imaging systems, although of course',
the patterns themselves would be quite different and would be made in pairs.
In general, image correlation devices are subject to the limiting
effects of two classes of image stress factors:
1. Image factors (frequency content, anisotropy, contrast) and,
2. Relief or terrain factors (slope, curvature, high order relief
distortions, and blind spots).
At present, image correlation systems can only be evaluated by using
photography of real terrain, which includes various combinations ofthese
stress factors in uncontrolled and often unmeasurable quantities. To perform
meaningful tests on the stereo scanner and to diagnose potential causes of
failure of correlation, it is necessary to use controlled test imagery which
contains clearly separated known amounts of these stresses. While it would
'46 impractical to devise test patterns for all the above stresses and
combinations of stresses in a limited program such as this one, certain
important items would be of interest.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For ease 2004/07/29 : CIA-RDP78B0477.00200010002-8
While a more detailed study should be made of the exact order of ,
importance of each stress, a reasonable set might consist of samples with
controlled slopes, samples with controlled curvatures, samples with higher
order terrain roughness, samples of flat terrain with variable anisotropy
(frequency content versus direction), samples with varying frequency
content, samples with varying contrast. In order that the number of
samples not be prohibitive, experimental parameters would be grouped about
values corresponding to actual photography which will be used in the
Automatic Stereo Scanner System.
WORK STATEMENT:
1. Analyze the types and magnitudes of image stress factors likely
to be encountered in the actual photography and specify a matrix
covering the likely combinations in the most economical. manner.
2. Devise methods for generating photographically a set of test
patterns as specified in 1.
3. Produce a sample set of test patterns for use on the Automated
Stereo Scanner.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For.lease 2004/07/29 : CIA-RDP7813047103002.00010002-8
TASK (3) - CORRECTION OF HIGHER ORDER DISTORTION IN SCANNING SYSTEM
Scan distortions presently u*d in the peak and valley correlator are
of six kinds: Zero order x and y translations and four first order distor-
tions, x and y scale, and x in y and y in x skews. These distortions can
be transformed exactly into optical counterparts, x and y parallax, differ-
ential zoom, differential rotation, and anamorphic distortion. These
optical counterpart distortions have exactly the same number of degrees of
freedom as the scan distortions used in the peak and valley correlator/ '
distortion analyzer. The quality of correlation is a direct function of the
amount of similarity between the two signals being correlated. Therefore,
if the signals to be correlated contain terms originating from higher order
distortions than those mentioned above, i.e. keystone distortion or terrain
curvature, these higher order distortions represent disagreements between
the two video signals and correspond to a kind of noise. Some of these
higher order distortions are relatively easy to generate electrically, but
have no optical counterparts; indeed only those transformations relating to
projective transformations can be removed optically.
While second and high-order(x- distortions that are present in the viewed
image are easily accommodated by the operator and do not degrade stereo fusion,
correlation reliability would be improved if these distortions were cleared in
the electronic scans of the image dissectors in the correlator. The value of
including these higher ordertenns may be investigated initially by simulation
on the PDP-1 computer, and then if positive results are obtaibe,c/l)by modifications
to the present video breadboard.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved Forilease 2004/07/29 : CIA-RDP78B0477.00200010002-8
WORK STATEMENT:
An investigation of this problem requires work in the following areas:
1. A theoretical study with two aims;
a) to predict the order of magnitude of improvement in
correlator signal to noise ratio when higher order
distortion terms are included.
b) to determine which distortions have the greatest effect
on correlator signal to noise ratio.
2. Preparation-of convergent and rectified pseudo convergent
(2 x 2 inch) slides for testtmaterial. Pseudo convergent
distortions is produced by rectifying the same picture with two
opposite projective distortions so that the effect of convergent
photography of perfectly flat terrain is produced. This eliminates
the higher order disturbances produced by terrain relief.
3. Design of a PDP-1 computer program for trapezoidal and/or quadratic
scans for correlation of convergent stereoscopic imagery.
4. Integration of trapezoidal and/or quadratic scan distortion circuits
into the present correlation system breadboard.
5. Correlation tests, preferably using material from Task 2.
6. Write a final report summarizing the findings of this investigation.
Approved For Release 2004/07/29 : CIA-RD1578B04770A000200010002-8
Approved For 410ease 2004/07/29 : CIA-RDP78130477000200010002-8
TASK (4) - SERVO STABILITY INVESTIGATION
The purpose of this task is to investigate the optical servo system,
particularly the anamorph servo loop to optimize the response time and
stability.
The basic instrumentation equations for the optical servo have already
been derived. (See Appendix 18 of Vol. II,"Project 9609 Sfudies and Tests".)
These equations indicate that the loop gain of the servo depends greatly on
the orientation of the two anamorphic lenses. In addition, singular points
occur when either 2 a = 0, 213 = 0, or 2(a - o where a and 8 are the
anamorph orientation angles. While the present system has been shown to
operate satisfactorily in the EROS breadboard, it is likely that an improve-
ment in stability and tracking speed could be obtained by further investigation.
The proposed investigation will be directed at analyzing the effects of
the above noted singularities on system performance and if necessary, modifying
the equations in order to obtain the maximum rate of change of anamorphism
consistent with system stability. The primary investigation will be mathe-
matical, but it is also proposed to make a practical test of the instrumen-
tation using EROS as a breadboard.
WORK STATEMENT:
1. Analyze operation of the anamorph servo system in the regions
a = 0, 0 = 0, and (a - = 0. Evaluate schemes for reducing the
effect of the singularities at these points an,8 maintaining constant
servo loop gain.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For "pease 2004/07/29 : CIA-RDP78B0477.00200010002-8
2. Design and assemble a breadboard servo system to be used with
EROS to evaluate the new control scheme.
3. Conduct a test program on EROS using photography in which the
anamorphic distortion lies near or crosses through the singular
points.
4. Write a final report covering the analysis, design and test
results.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved Foreease 2004/07/29 : CIA-RDP78130477000200010002-8
TASK(5) - CORRELATION WITH PARTIALLY OBSCURED IMAGES
At present if half of the field of view in one member of a conjugate
pair is obscured by clouds or a frame edge, the Electronic Correlator cannot
operate. This is because the first order distortion error signals depend on
receiving valid information from both halves of the field of view. However,
it may be possible in this case to operate with only zero order (parallax)
corrections, and thus to stay locked in.
A proposed technique involves using four disagreement summers instead
of one for determining correlation thresholds to insure correlation on two
halves of the field of view before closing the distortion loops. The
parallax loops would be closed if correlation were present in any quadrant
of the field of view.
Another technique for approaching an edge would be to apply parallax
shifts to both rasters in the same direction (away from the edge) such that
while the eyepiece field of view would include the edge, the rasters would be
offset to only look at the area of good correlation.
These techniques can be investigated with the image dissector breadboard.
WORK STATEMENT:
1. Investigate techniques for maintaining correlation when approxi-
mately half of the scanned area is obscured or otherwise uncorre-
latable.
2. Design and assemble modifications to breadboard scanning system to
evaluate these techniques.
3. Write a report summarizing the findings of this investigation.
?
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For geese 2004/07/29 : CIA-RDP78130477.00200010002-8
TASK (6) - SEARCH SIMULATION
The purpose of this task is to use the PDP-1 correlation simulation
program to develop a search strategy to be implemented in the Stereo Scanner.
The following process is used to obtain estimates of distortion settings.
A point is selected in the master frame and its coordina;ps determined% These
coordinates are used in the camera system model with the data pertaining to the
geometry of the exposure to project this point to the datum and thence, up to
the second exposure station. The zero and first order coefficients of this
transformation are computed and instrument settings determined. The instrument
Is set to these values and correlation is attempted.
The error sources in this process are (1) the coordinates in the master
frame are imperfectly known, (2) the data of the taking situation contain errors,
(3) the model does not account for relief and hence the coordinates of the
"conjugate" are in error, (4) there are, finally, errors in setting the instru-
ment. All of these combine to introduce errors in the estimated distortion
settings.
The magnitude of the contribution from each of these sources must be
evaluated so that search routine requirements can be specified. It should be
noted that as there are six distortions to be corrected, the search must be
made in a mathematically six-dimensional space.
In actual practice, should the computer position the optics outside of the
pull in range of the correlator, the operator will be able to clear parallaxes
and fuse the images. The search hlutine will eliminate the need for human
intervention except in the most extreme error conditions when a search would
consume too much time.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved Foripease 2004/07/29 : CIA-RDP78130477000200010002-8
To effectively implement such a search we must determine those para-
meters and their thresholds which indicate existence or imminence of corre-
lation lock-on, and those system characteristics which indicate likely trouble
conditions.
Finally, a search strategy must be defined which includes defining
search increment size, region to be searched, configuration of search and
policy alternatives.
SinCe we will have little time to experiment with decision schemes once
the control computer is mated to the stereo scanner; we should gather as much
information as possible from a computer simulation of the search process.
A series of experiments should be constructed utilizing this simulation, which
will yield answers to the questions raised above, and which will lead to the ,
design of a workable search strategy.
WORK STATEMENT:
1. Evaluate magnitude of expected position and distortion errors in
setting up the stereo scanner.
2. Devise suitable search routines, taking advantage of any systematic
associations between the various errors to reduce the volume of the
space or the dimensions that must be searched.
3. Program the PDP-1 computer for the most suitable search routines
and evaluate them under varying conditions.
4. Write a report summarizing the work and the results achieved.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For/lease 2004/07/29 : CIA-RDP78130477.00200010002-8
TASK (7) - DETERMINATION OF THE OPTIMUM SEQUENCE FOR DISTORTION REMOVAL
The PDP-1 correlation tests performed to date indicate that it is
desirable to reduce the zero order x and y translation errors to essentially
zero before operating on the first order distortion errors. Further tests
indicate that among the first order distortions there is, for any given image,
an optimum sequence in which errors should be removed. Results so far are
inconclusive as to whether the optimum sequence for one image provides an
optimum sequence for a wide range of input photography.
It is anticipated that as photographs with more and more relief are
inserted into the system, the correlation reliability will decrease, but that
significant improvements may be possible if an optimum sequence can be
determined.
We are, therefore, proposing a test and data analysis program using
existing PDP-1 computer programs and techniques, on photographs containing
varying degrees of relief. The purpose of the investigation will be to deter-
mine the behavior of the correlation system and to arrive, if possible at a
sequence for distortion removal which will optimize correlation reliability.
WORK STATEMENT:
1: Using the existing PDP-1 computer program for distortion correction,
investigate the rate of correction of the six distortion parameters
with stereo pairs containing varying amounts of vertical relief.
2. Investigate the effect of different distortion correction sequences
on the speed and reliability of the distortion correction process.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For
ase 2004/07/29 : CIA-RDP781304770.0200010002-8
3. Determine, if possible, an optimum sequence that is effective
over the greatest range of input material.
4. Write a report summarizing the findings of this task.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For ease 2004/07/29 : CIA-RDP78130477.00200010002-8
TASK (8) - INVESTIGATION OF MAGNITUDE OF UNCORRECTED SECOND ORDER Y-PARALLAXES
An optical distortion correction system consisting of zoom lens, image
rotator, and anamorphic lenses can correct only first-order distortions. In
convergent photography second-order systematic distortion are present, as
well as higher order distortions due to terrain relief. Distortions in the x
direction of the second- and higher-orders can be accommodated visually in the
stereo model and do not present a problem. However, uncorrected Y-parallaxes
do present a potential viewing problem, as these distortipns do not correspond
to any realizable terrain model. Correction of second- arid higher-order
distortions in an optical viewing system is difficult to implement and should
only be undertaken if absolutely necessary. We therefore, propose to investigate
the magnitude of residual second- and higher-order Y-parallaxes in the various
photographic formats to be used in the stereo scanner.
The results of this investigation will determine whether uncorrected Y-
parallaxes do in fact constitute a problem in stereo viewing, and will indicate
whether work on optical second-order correction is necessary.
WORK STATEMENT:
The formats of the three photographic systems to be used in the stereo
scanner will be examined and the magnitudes of the uncorrected Y-Parallaxes for
the worst cases of practical interest, will be calculated. These results will
be compared with the values of acceptable Y-parallax already determined and
published in the literature.
A conclusion will then be drawn on the desirability of attempting to
correct second order Y-parallaxes in future stereo viewing equipment.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Approved For *ase 2004/07/29 : CIA-RDP7860477
SCHEDULE
0200010002-8
The duration of all proposed tasks is four months with the exception
of task 2 which is seven months.
Tasks 1 through 5 will be commenced simultaneously at the start of
the contract. To obtain efficient utilization of available manpower and
the PDP-1 computer, the starting dates of tasks 6 and 7 and 8 are
staggered as shown in the bar chart.
All tasks will be completed within eight months.
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
ase 2004/07/29 : CIA-RDP78B047700)0200010002-8
?
I nrigilMEMBEINIENIEHRgraa MINEIBMINNESIBEI
COMMINIIIMMISIIIIILIBIESIECIE91111111111111111111=11111113?Taniiiiii
t
r
rt
+-Ft- 4
L
4,1
L74-
4
um 11 ma
MHO
LE EL . no :wuneolernike
I.
? lin:
nano
sans
man
11110 ii
:
r
.... ... .
o ..
... ..
?
".ron..ur 1
Rupiah
ini.m.! nom .1.1iiir hpidno:.
1 lig .1.111
....
1., mu ihihn
n nu Hi "rift :I I
. FR .ohLiii ..: thi . sobs
k 1 hyldnummung 1 so
MUM "Mar 11111.1111 FiLigladh g
MI i ...0-11Totril
irpriiiiiiii:00 Nikai gliollin .1 0 : isquitirl -0-
. MI III MN II a M Bain jallEdi
HthdlOPPIIPOrill MIUMUILI 111111MIlin
Slitl MILT . wariaigs..." woo
1119111nribirldll ma gluing rat: r IrraNe NH
minim ?? ????? ?__ ?? ? b-.1... -.ILAN psi
? ? w????? ?? itan?laamiono
?
RIZ; Ern!IMPI*_?.?...? _ . gilirMr1P.,.,,i!rilL.VIi! 9MM Pm' TWO:
- 1.-- .., -pi
- .
1
Ann L
lull a WillMAL iv
PUIIUUPRUIUUIUP kr um
1 us
... MMMMMM mon
solimah ow hil
twqmpirs in.i mmq
. . Aldluirool TIP.,
uk
a ii a ign nip 1 um n llid
I Ibur?...4.:7C.......;:liiiium6hilkonda
STAT Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8
Next 7 Page(s) In Document Exempt
Approved For Release 2004/07/29 : CIA-RDP78604770A000200010002-8