PROPOSAL FOR DEVELOPMENT OF CHANGE-DETECTOR EQUIPMENT
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP78B04747A002600010007-6
Release Decision:
RIPPUB
Original Classification:
K
Document Page Count:
70
Document Creation Date:
December 28, 2016
Document Release Date:
December 3, 2001
Sequence Number:
7
Case Number:
Publication Date:
January 20, 1962
Content Type:
REPORT
File:
Attachment | Size |
---|---|
CIA-RDP78B04747A002600010007-6.pdf | 3.08 MB |
Body:
Approved FoRelease 2002/01/02 : CIA-RDP78B047002600010007-6
STATINTL
Copy No. _..+
PROPOSAL FOR DEVELOPMENT OF
CHANGE-DETECTOR EQUIPMENT
20 January 1962
STATINTL
Declass Review by NIMA / DoD
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
This data shall not be disclosed outside the Government
or be duplicated, used or disclosed in whole or in part
for any purpose other than to evaluate the proposal, pro-
vided that if a contract is awarded to this offeror as a re-
sult of or in connection with the submission of such data,
the Government shall have the right to duplicate, use, or
disclose this data to the extent provided in the contract.
This restriction does not limit the Government's right to
use information contained in such data if it is obtained
from another source.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Page
II CONCEPT OF CHANGE DETECTION . . . . . 3
1. Strategic and Combat Considerations 3
2. Military Applications . . . . . . . . . . . 5
3. Comparison of Various Methods of Change
Detection . . . . . . .. . . . . . . . . . 8
III RECOMMENDED SYSTEM . . . . . . . . . . . 17
1. System Considerations . . . . . . . . . . 17
2. Image Registration . . . . . . . . . . . . 19
3. Data Comparison . . . . . . . . . . . . . 29
4. Rejection of Unwanted Data . . . . . . . . . 32
5. Results of Present Studies . . . . . . . . 35
6. Growth Potential . . . . . . . . . . . . . 40
PROPOSED DEVELOPMENT PROGRAM . . . . 42
1. Design Objectives . . . . . . . . . . . . . 42
2. Equipment Description . . . . . . . . . . . 43
a. Components . . . . . . . . . . . . . 43
1. Operational 'Sequence . 43
c. Operator Options . . . . . . . . . . . 47
3. Task Descriptions . . . . . . . . . . . 48
a. General . . . . . . . . . . . . . . 48
b. Task I - System Predesign . . . . . . . 49
C. Task II - Display Data Processing
Studies . . . . . . . . . . . . . . . . 50
d. Task III - Design and Fabrication . . . 51
e. Task IV - Checkout and Evaluation . . . 59
T. Task V - Support Engineering . . . . . 61
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fot elease
TABLE OF CONTENTS
Page
4. Organization and Schedule . , . , . . . . . . . 63
a. Project Organization . . . . . . . . . . . 63
S, Program Schedule . . . . . . . . . . . . 63
Personnel . . . . . . . . . . . . . . . 63
Appendix
A RESOLUTION CALCULATIONS . . . . . . . . . . 73
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Figure
Page
1 Detection of Vehicle Movement . . . . . . . . . . . 7
2 Detection of Mines . . . . . . . . . . . . . . . 9
3 Time Required for Various Methods of Change
Detection . . . . . . . . . . . . . . . . . . . . 1 1
4 Omission Errors for Various Methods of Change
Detection . . . . . . . . . . . . . . . . . . . . 12
5 Commission Errors for Various Methods of Change
Detection . . . . . . . . . . . . . . . . . . . . 14
6 Efficiency of Various Methods of Change Detection . . 15
7 Manpower Requirements for Side-by-Side and
Difference Methods . . . . . . . . . . . . . . . . 16
8 Simplified Block Diagram of Change Detector . . . . 18
9 Typical Display with Various Changes . . . . . . . 20
10 Correlation of Two Scenes . . . . . . . . . . . . . 21
11 Three-Dimensional Correlation Surface . . . . . . 23
12 Generation of Error Signals . . . . . . . . . . . . 24
13 Autocorrelation Curve . . . . . . . . . . . . . . . 26
14 Detector Signal during Search Operation . . . . . . 28
15 Static Match Curve for Magnification . . . . . . . 30
16 Error Signal for Magnification . . , , , . . , , 30
17 Block Diagram of Proposed Change Detector. . . , 31
18 Operation of Video-Difference Detector . . . . . . 33
19 Simplified Block Diagram of Video-Difference
Detector . . . . . . . . . . . . . . . . . . . . 33
20 Transmissivity of Shadow Areas . . . . . . . . . 34
21 Two Change-Detection Methods. . . . . . . . . . . 38
22 Console Design . . . . . . . . . . . . . . . . . . 44
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fot%Release 2002/01/02 : CIA-RDP78B047002600010007-6
Figure Title
Page
23 Control Panel . . . . . . . . . . . . . . . . . . 60
24 Project Organization . . . . . . . . . . . . . . . 64
25 Program Schedule . . . . . . . . . . . . . . . . 65
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fotielease 2002/01/02 : CIA-RDP78B047002600010007-6
SECTION I - INTRODUCTION
STATINTL
STATINTL
No
The photo-interpreter in an information processing system must search a
photograph, compare it with other photographs of the same area and with
surrounding areas, and compile data that will finally be fitted together care-
fully for an appraisal of the strategic, tactical, and.logistic significance of
any changes that have occurred between the photographs.
Screening aids offer the greatest potential for the improvement of photo-
interpreter performance in the analysis of surveillance data. These devices,
while leaving the job of interpretation to the human, provide a machine meth-
od to reduce greatly the search time of the photo-interpreter; thus, they en-
able him to handle large amounts of data in a given time or reduce the time
required to obtain a given piece of intelligence.
A device that will automatically register, compare, and display data from
two perspective views of a common area taken at different times exemplifies
a screening aid that will assist the photo-interpreter. Such a device allows
the interpreter to locate changes between scenes quickly and, subsequently, to
determine the nature of the change (for example, the addition of new objects
to the scene, the change in positions of objects within the scene, and the re-
moval of objects from the scene). A change detector screening aid also pro-
vides a basic device around which other screening aids can be added to im-
prove performance further.
tnis. pro-
gram comprises the design, fabrication, and evaluation of a change-detector
screening aid that will provide the capability described above. The equip-
ment will consist of a console containing comparator and display units. The
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved FoIease 2002/01/02 : CIA-RDP78B047002600010007-6
comparator unit will contain the necessary components to perform the func-
tions of automatic registration and comparison of image data from roll film.
The display unit will provide the components required to display the scenes
being compared as well as the output comparison data and controls required
to operate the console. Described herein are some of the operational uses
of the device, the basic technique proposed, and a summary of recent studies
and problems that remain to be solved.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Fo lease 2002/01/02: CIA-RDP78B04744002600010007-6
Surveillance is conducted primarily to determine the location of the
enemy and chart his movements. Today, the development of an adequate
strategic and combat surveillance capability has become a serious prob-
lem confronting the Military because of the advent of missiles, satellites,
and nuclear warheads together with the development of innovations in
tactics and strategy created by the use of the new weapons.
Prior to and during any war, strategic intelligence must be gathered
concerning the capabilities, limitations, vulnerabilities, and probable
actions or reactions of foreign or enemy nations. Satellites are now STATINTL
used and will continue to be used for surveillance. Intelligence concern-
ing topology and military geology can now be obtained by the use of visual
sensors. In the near future, technol-
ogy will be available to permit the use of such sensors in satellites and
very high altitude aircraft. The imminence of hostilities, for example,
can be measured by comparisons of data obtained from repeated looks at
large geographical areas.
As a global satellite surveillance system is developed and as space flights
become more frequent and routine, there will be an increasing amount of
strategic surveillance data available for interpretation. Automatic de-
vices are required that will screen and therefore reduce the amount of
data to be interpreted. The key to this process is a technique that de-
tects the positions within pictures where changes have occurred when
compared to pictures taken previously of the same area. In some in-
stances, where the low resolution of pictures makes detail interpretation
impossible and would normally obscure changes, this technique will
reveal general areas of change. Subsequent reconnaissance missions
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
r Approved Fotklease 2002/01/02 : CIA-RDP78B0474iA002600010007-6
can then be performed to obtain more detailed pictures from which
specific changes can be determined.
During general or limited war the combat surveillance problem is be-
coming increasingly important. Timing has become critical since armies
can now move rapidly and contain integral capabilities for air transporta-
tion. Targets now move constantly, employing camouflage, night, or
bad weather to avoid detection. Thus, the timely location of targets and
the resulting shift of the defense posture require quick decisions and
actions. This can be accomplished only by more rapid, complete, and
precise intelligence.
Present tactics and organization require a dual capability for the Army,
namely, strategic and tactical mobility. The role of tactical aviation
for close support will become partly the responsibility of surface-launched
missiles and surveillance systems organic to the ground forces. All sur-
veillance functions have become more difficult because of the high mobil-
ity possessed by the enemy. This mobility will continue to increase with
time, hence profitable targets will remain profitable only for short peri-
ods. The total enemy situation observed or sensed at any given time may
change in minutes. Even now, the existing data-gathering capabilities
exceed the capacity to process this information. The systems currently
under development are planned to gather larger amounts of data, making
the problem truly difficult. The information-processing system must be
built to exploit the capabilities of both men and machines. The system
must be matched at one end with the data-gathering capability of surveil-
lance and other intelligence sources and at the other end with the basic
information required for a command decision.
Dr. Kraft estimates, for example, that the field army intelligence re-
quirements in a highly mobile war may dictate the data processing of as
high as 100, 000 pictures daily. a Present estimates of interpretation
aDr. Conrad Kraft, "Possible Solutions for Minimizing Time Delay and
Errors in Data Utilization. " Proceedings of the National lAS/ARMY Avia-
tion Meeting (UNC). April 13-14, 1961, Washington D. C.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Foiklease 2002/01/02 : CIA-RDP78B0474 AO02600010007-6
rates indicate that as many as 12, 000 man-years of work would be re-
quired to analyse this amount of film for a years' combat.
The data interpretation of imagery in these amounts is well beyond the
capacity of the trained personnel available to the United States. The
training of large numbers of interpreters alone is not an adequate solu-
tion to this problem. The solution will be found only in the utilization of
better methods of employing machines to aid photo-interpreters in han-
dling large amounts of data while leaving the final integration of the data
to the photo-interpreter,.
Machines can be built that possess capabilities greater than man in
memory, data handling, and the rapid transfer of data. Thus, a simple
and direct approach is the utilization of semitrained personnel to operate
simple but reliable equipment that can detect rapidly and accurately any
changes that have occurred in two views of a common area taken at d_.'
ferent times. This approach permits the skilled, highly trained photo-
interpreter to concentrate on the details of the changed areas, where the
highest probability of significant data exists.
Comparisons of repeated looks of a common area reveal three types of
image changes: (1) the addition or movement of objects into the area,
(2) the removal of objects from the area and, (3) the movement or re-
positioning of objects within the area, Knowledge of the types of changes
with identification of the objects concerned yields intelligence for a
variety of military situations pertaining to enemy capabilities, limita-
tions, and vulnerabilities. The measure of transportation and other
activities is useful as an indicator of forthcoming action.
Activity can be defined, in general, as the number of changes taking
place in a given area of interest. Thus, counts of changes per unit area
constitute a measure of activity. However, the changes measured must
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved FoRelease 2002/01/02 : CIA-RDP78B047A002600010007-6
SECTION II - CONCEPT OF CHANGE DETECTION
STATINTL
be of military significance if the intelligence derived from such measures
is to be useful in military decisions.
The imminence of hostilities for a general war can be measured by
examination of enemy activity over large areas. Supply area activity,
the location of new construction, and the relocation of military targets
can all be used to predict the coming of a general war. A change detec-
tor can be used to screen the vast amounts of data required to obtain an
over-all imminence-of-war indication rapidly enough to permit defensive
actions.
Enemy intentions toward a limited war can be judged from changes of
such items as the image background and motion. The most reliable in-
dication of a danger situation can be provided by a combination of the
parameters of motion appearing simultaneously, or in a logical sequence,
starting with slower motions (in central or remote areas); followed by
a convergence toward areas where an order of battle is being formed for
the instant of strike. A reliable danger count would be a summation of
changes as they appear simultaneously or as new changes are added to
those already existing. Figure 1 exemplifies the use of a change detec-
tor to detect vehicle movement within an area.
The continued observation and subsequent measurement of changes that
have occurred in the terrain itself can be used to locate new bunkers,
munition dumps, camouflaged areas, and new missile launching sites.
For quantitative bomb damage assessments a change detector can be
used to compare the reconnaissance pictures taken before and after a
strike. Repeated comparisons of detail pictures of a combat area can
be used to plot an enemy order of battle since the movement and re-
location of various military targets, personnel, and the construction of
foxholes and bunkers can be quickly located and recorded.
The use of a change detector to locate the position of newly laid mine
fields is a promising concept. In many cases where great effort has not
-6-
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
SECTION II - CONCEPT OF CHANGE DETECTION
NEW VEHICLE
POSITIONS
urc I - Detection of Vc?hiclc Mover -nt.
STATINTL
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved Fc,elease 2002/01/02 : CIA-RDP78B04 4-002600010007-6
STATINTL
been expended to hide the mines, a simple comparison in a change
detector of the area at different times will easily detect the addition of
the mines. Figure 2 shows a typical example of the use of a change
detector for the location of newly laid mines. For those cases where
previous reconnaissance pictures are not available or where great care
has been taken to hide the mines, the use of camouflage detection film
with a change detector can be used to locate spoil lines left by the burial
of the mines themselves or to locate differential growth in the mined
areas. Processed camouflage detection film shows highly infrared-
reflective objects as red (e. g. , healthy foliage), green objects that are
not highly infrared reflective as blue-green (e. g. , unhealthy foliage,
which, although still green, has lost most of its infrared reflectivity),
and brown or red objects;, which are not highly infrared reflective, as
yellow or brown (e. g. , the dead foliage of most plants).
Sod covering mines, for example, will experience healthier growth than
the surrounding area, while patrol or egress paths through a mine field
will normally scar the vegetation, causing a decline in growth. Thus,
pictures of mine fields made with camouflage detection film, taken even
after the planting of mines, will exhibit the differences in growth so that
the normal identifying mine field patterns can be observed.
There are three basic methods of detecting changes between two scenes
of overlapping data. The first, the side-by-side or juxtaposed method,
is the current operational technique and consists of the intermittent
scanning of two images of the same area placed side-by-side in front of
the observer. The pictures are scanned alternately until the observer
recognizes that data exist in one scene and not in the other. He must
alternately store and compare as much data as he is capable of handling.
The second method is the apparent motion, or flicker:, method, since any
change occurring in the scene will have an apparent motion or flicker to
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6 - 8
STATINTL
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
SECTION IT - CONCEPT OF CHANGE DETECTION
CHANGE DISPLAY SHOWING MINE PATTERNS
Fi['ure Z - Detection of Mines
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved For-,&Iease 2002/01/02: CIA-RDP78BO474 02600010007-6
the observer. The method consists of the presentation of old and new
images in registry and in alternating sequence at approximately 1-1/2
cps.
STATINTL
The third technique is the "difference" or "overlay" method and consists
of the optical or electrical mixing of positive and negative images of the
old and new scenes. The optical method produces the quotient of the two
transparencies,, while the electrical method normally produces the dif-
ference between the two images.
The three methods were compared for various types of changes and
Changes in the number, size, and position of targets were independent
variables while (1) the time required to detect the changes and (2) the
errors in the changes were the dependent variables. Fifteen trained
observers were used on each of the three methods. The number of
false reports of targets that were not changed were determined.
The differences between the three methods proved to be highly signif-
icant. Figure 3 shows the average time required to locate as many as
six changed objects within a scene for the various methods of change
detection; the juxtaposed method is approximately three times longer
than the difference or flicker method. The averages for all groups of
changes are 58. 5, 19. 7, and 22. 5 sec, respectively. Although the
display size did not appear to affect the detection time of the side-by-
side method, it affected significantly the difference and flicker methods.
Figure 4 shows the omission errors for the various change detection
techniques. The data indicate that positional changes are the most dif-
ficult to detect. For the side-by-side method as many as 99-percent
errors can be expected with change targets of four-minute size with no
display magnification. The error percentage can be reduced to 60 and
37, respectively, when display magnifications of 4 and 8 are used. When
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved For, please 2002/01/02 : CIA-RDP78BO474W02600010007-6 STATINTL
SECTION II - CONCEPT OF CHANGE DETECTION
NOTE
TARGET DISPLAY SCALE = 4 MINUTE UNITS
AVERAGE REPORTING TIME = 8 SECONDS
NUMBER
SIZE
POSITION
Figure 3 - Time Required for Various Methods of Change Detection
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
0
50
4
DISPLAY MAGNIFICATION
TYPES OF CHANGE
STATINTL
Approved For'Q'lease 2002/01/02 : CIA-RDP78B0474002600010007-6
`?
SIDE-BY-SIDE
OVERLAY
\ ?
APPARENT MOTION
1 4
DISPLAY MAGNIFICATION
TYPES OF CHANGE
NUMBER
SIZE
? POSITION
Figure 4 - Omission Errors for Various Methods of Change Detection
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved FoN$elease 2002/01/02 : CIA-RDP78BO474JJM02600010007-6
changes are caused by the addition of targets, the display size has little
effect. The difference and flicker methods however, have minimum
omission errors when a display magnification of 4X is used.
The number of falsely detected changes, called commission errors, are
shown in Figure 5, The efficiency of the various methods can be obtained
by combining the detection times and the errors as shown in Figure 6.
For the experiments performed by Dr. Kraft the difference and flicker
methods were essentially the same.
Consider a typical global satellite surveillance system, operating at an
altitude of 150 naut mi and with a 24-in. -focal-length camera, designed
to obtain a complete mapping of the USSR and China every 30 days. Such
a system will generate at least 46, 500 pictures each month. If a cloud
cover of 0. 66 is assumed, the number of frames will have to be in-
creased to 140, 000 pictures.
Consider a field-army battle-front situation with daily reconnaissance of
an area of 100 by 300 naut mi from an altitude of 2, 000 ft using a 9-in.
film format and a camera focal length of 12 in. This reconnaissance
will produce approximately 480, 000 frames per day. The relative man-
power requirements to determine the changes in the scenes during a 24-
hr period each day are shown in Figure 7 for the side-by-side and the
difference change detection techniques. Approximately 250 men would
be required for the present side-by-side technique and still 33 percent
of the changes would go undetected. If the difference detection technique
were used, only 54 men would be required and only 3. 2 percent of the
targets would go undetected. The number of errors could be reduced
further by use of a flicker technique in conjunction with the overlay
method.
The experiments performed to date have been somewhat limited since
they were made in the noise-free condition. The difference between the
conventional side-by-side and either the difference or flicker techniques
is sufficiently large to indicate that the latter methods would greatly im-
prove operator performance.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Fo elease 2002/01/02: CIA-RDP78BO474ZC002600010007-6
SECTION II - CONCEPT OF CHANGE DETECTION
W
0
60
SIDE-BY-SIDE
30
20
7
J
0
10
M
N
0
O
30
W
Z
20
O
U)
N
10
DISPLAY MAGNIFICATION
TYPES OF CHANGE
NUMBER
SIZE
POSITION
NOTE
TARGET DISPLAY SCALE = 4 MINUTE UNITS
Figure 5 - Commission Errors for Various Methods of Change Detection
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved Foi:$elease 2002/01/02 : CIA-RDP78BO474Z 002600010007-6 STATINTL
SECTION II - CONCEPT OF CHANGE DETECTION
0
90
NUMBER
SIZE
POSITION
NOTE
TARGET DISPLAY SCALE = 4 MINUTE UNITS
AVERAGE REPORTING TIME = 8 SECONDS
Figure 6 - Efficiency of Various Methods of Change Detection
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved Foelease 2002/01/02 : CIA-RDP78B0474,#002600010007-6
STATINTL
Figure 7 - Manpower Requirements for Side-by-Side and Difference Methods
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
The proposed system is essentially a difference or overlay type change
detector. Since the registration of the two images is also required for
the flicker method, this equipment will have the optional capability of
providing such a display to the viewer. The electrical mixing of scene
data was chosen because of recent studies (see item 5, this section).
Equipment is proposed that will automatically register and compare two
images and then read-out and display the differences between them.
Figure 8 shows a simplified functional block diagram of the change de-
tector. The first operation in the change detection process is the selec-
tion of two images (referred to as the comparison scene and the refer-
ence scene) of the same general area in which changes are to be detected.
The two selected images are inserted in the change detector and scene
information is brought into registration by means of an automatic cross-
correlation technique. Registration normally requires four degrees of
freedom namely x, y, 0, and magnification. Translational displacement
(x and y) of one scene relative to the other is necessary, since it is very
unlikely that the centers of the two scenes will have the same geographic
location. The rotation of one scene relative to the other is necessary to
compensate for any differences in the azimuth alignment of the images.
The magnification of one scene must be adjusted to compensate for dif-
ferences in the image scale-factors that result if camera focal lengths
differ or if the pictures were obtained from different altitudes.
After the two scenes have been properly registered, they are subjected
to a comparison process whereby each scene is examined in detail and
a comparison is made resulting in a display signal proportional to the
difference between the two scenes.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Foi elease 2002/01/02 : CIA-RDP78BO47 002600010007-6
COMPARISON
SCENE
REGISTRATION 1
PROCESS SCENE INPUT
COMPARISON AND
DIFFERENCE
DETECTOR
DISPLAY
PROCESSOR
CHANGE
DISPLAY
Figure 8 - Simplified Block Diagram of Change Detector
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
No Approved Fo?Celease 2002/01/02 : CIA-RDP78BO474UO02600010007-6
A signal is displayed that is proportional to all the differences between
the input and reference scenes. While this display permits quick identi-
fication of those locations where changes have occurred between scenes,
it is not limited to changes that may have military significance. Exam-
ples of undesired changes are (1) the existence and orientation of shad-
ows, (2) cloud cover, and (3) system noise. It is therefore desirable to
reject (filter) these differences, which have for the most part little
military significance (see item 4, this section). Figure 9 shows a
typical display with various changes noted. The coordinates of the dis-
played changes with respect to the reference scene can be obtained by
inclusion of a reference coordinate system that can be superimposed on
the monitor at the time of readout.
To accomplish registration, one of the images is moved to align its
scene objects with the corresponding objects on the other image. The
two images must be brought into register accurately before the changes
can be detected. For example, if the movements of vehicles or construc-
tion of small buildings must be detected, the registration error must be
less than the dimensions of these objects (i. e? , five feet or less). This
process is performed automatically through a correlation technique.
Although various methods are available to correlate image data, the
theory of operation is essentially the same for each method. Consider,
for example, that the comparison scene consists of a positive transpar-
ency that is back-lighted as shown in Figure 10. This scene is imaged
on the reference scene, which is also a positive transparency. - The
light transmitted through the reference transparency and focused on a
photo-sensitive element is a measure of the crosscorrelation between
the two images.
When the two scenes are correlated, light from the information in the
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
NEW SHIP
POSITION
SHADOW'; AND
PERSPF -T',VE
cHANGFi
PiANGE DISPLAY
irF t - I'ypYcal Display with Various C' in -f,s
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved Fo?eIease 2002/01/02 : CIA-RDP78B0474ZA002600010007-6
W
z
w
0
zz
ow
o J
W
J
F
0
W(n
-) Z
o W
0J
Figure 10 .- Correlation of Two Scenes
611
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
STATINTL
Approved Fo lease 2002/01/02: CIA-RDP78B0474Zj1002600010007-6
input scene is directed on corresponding information in the reference
scene, which results in a maximum of transmitted light..
The light transmitted through the reference transparency is proportional
to the product of the transparency functions of the input and reference
scenes. Mathematically, the process can be defined as
L = JA[c(x,y)1 [~rk + a, y + /3)]dA ,
= transparency fundtion'of the comparison scene,
c
r = transparency function of the reference scene,
x and y = ground coordinates of each target in the scenes,
CY and /3 = relative displacements between the two scenes,
A = area of the comparison scene.
Figure 11 is a three-dimensional representation of the light detected at
the phototube after it passes through the reference transparency as the
reference is moved in directions CY and /3 relative to the comparison
scene. An error signal is generated that is used to drive the reference
scene in the tx and /3 directions so that the detected light flux is max-
imized. The process is essentially the same as that described math-
ematically by the correlation function.
If one of the scenes is a positive transparency and the other a negative
transparency, the point of maximum correlation is represented by a
minimum of transmitted light.
Figure 12 shows a two-dimensional correlation curve for a positive and
negative transparency that results when the reference and comparison
scenes are displaced in one direction only. This curve is normally
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved For"elease 2002/01/02: CIA-RDP78B0474Za(,Y'002600010007-6
Figure 11 - Three-Dimensional Correlation Surface
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Forl4elease 2002/01/02 : CIA-RDP78B0474Z 002600010007-6
A. GENERATION OF ERROR SIGNAL
r SIN wt
P = da 0
r = r SINot
0
Z2
+aL
Figure 12 - Generation of Error Signals
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6 -24
Approved Forelease 2002/01/02 : CIA-RDP78B047t002600010007-6 STATINTL
SECTION III - RECOMMENDED SYSTEM
referred to as the static match curve.. A direction-sensitive error signal
can be generated from the correlation signal curve by nutation of the
reference scene relative to the comparison scene. In Figure 12, the
nutation is a sinusoidal displacement in the CY direction and produces a
change in transmitted light that, in turn, produces a varying phototube
output. The phototube output together with a nutation reference signal is
fed into a phase discriminator, which generates a signal proportional to
slope of the correlation curve.
If the selected nutation radius, ro is small enough, the portion of the
curve from a to b can be considered a straight line. Then the output of
the phototube will be
P dcx ro sin Wt
and the output of the phase discriminator will be proportional to dL/dx.
This signal can then be used to determine the direction of the mismatch
and to drive the transparencies to the point of best correlation (match
point). A plot of this error signal as a function of mismatch, CL, is
shown in Figure 12, Figure 13 shows an aerial photograph and the re-
sulting two-dimensional correlation curve obtained when the photograph
is correlated with itself.
Figure 12 shows that an error signal can be obtained with scene dis-
placements between -01L and a . This value of displacement is called
the lockon range of the correlator. If the initial displacement between
the two scenes exceeds this range, a search operation must be imple-
mented to move one scene relative to the second scene in such a pattern
that, at some time during the search, the film will pass within the lock-
on ranges thus generating an error signal. The output of the phototube
is monitored by a BBmatchpoint10 detector that determines by the estab-
lishment of a signal threshold, when the two scenes are near the match
point. When this occurs, the search operation is stopped and the system
is driven into exact registration by means of the error signal. An al-
ternate implementation consists of a search through the complete pattern
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
SECTION III - RECOMME;NDEO SYSTEM
6 t 1 1 1 ) )-
-DISPLACEMENT (FEET)
'i,ri)rt3 - Cum lation Curve
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fa elease 2002/01/02 : CIA-RDP78BO474 002600010007-6
and storage of those coordinates that result in the greatest correlation
signal. The error signal is then used to return the displaced scene to
the stored coordinates for exact registration. This alternate approach
will be used since it does not require the establishment of a signal
threshold and is therefore relatively insensitive to changes in film base
densities and contrast. A typical example of a detector signal obtained
as a function of displacement in a search operation is shown in Figure
14. The upper trace is a record of the search position versus time in
one direction of an Archimedes spiral search pattern while the lower
trace is the input signal to the match-point detector.
Correlation in the azimuth direction can be performed in a manner
similar to that just described for the x-y direction. If one scene is ro-
tated relative to the other, a two-dimensional correlation curve similar
to the one shown in Figure 12 will be generated. Application of azi-
muthal nutation and phase-discrimination, results in the generation of an
error signal 'similar to the one shown in Figure 12.
If the scale-factors of the two scenes are not the same, the correlation
between the two scenes can be described by
fA
= reference scene,
r
c = comparison scene,
x and y = coordinates of reference scene,
x' and y' = coordinates of comparison scene,
a and a = relative displacement between the two scenes,
u = scale factor.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
x, yJ [c( ux' + a, -Zy' + j31 dA ,
STATINTL
SECTION III - RECOMMENDED SYSTEM
Approved Fo??Release 2002/01/02 : CIA-RDP78B047002600010007-6
V-VT 11\\11\1\\1\11\\111\\\1\1
c
/77/
-Uh
Figure 14 - Detector Signal during Search Operation
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
iJ?}~rr1
STATINTL
Approved FoV4&elease 2002/01/02: CIA-RDP -
SECTION III - RECOMMENDED SYSTEM
If c and 0 are constant, this function can be represented by a static
match curve as a function of ,i. Nutation can also be applied to the
magnification matching process and an error signal generated in the
same way as described for the case where the scenes are displaced in
x and y. A scale-factor nutation amplitude of one percent of the normal
value of scale factor is normally adequate for correlation. Nutation in
x, y, and -~ can be performed simultaneously if the nutation frequencies
chosen are sufficiently different to permit their separation by filtering
with a minimum of cross talk.
The correlation curve (static match curve) for various scale-factors
obtained from an aerial photograph is shown in Figure 15. The error
signal generated by nutation and subsequent discrimination is shown in
Figure 16. This signal is used to close a magnification (scale factor)
loop, thus adjusting the scale of one scene to correspond to the scale of
the other scene. Figure 17 is a detail block diagram which shows the
various functions required to provide registration in x and y, azimuth,
and magnification.
When the comparison and reference images are registered the scene
data are compared to determine whether or not any differences or
changes exist between the two images. This is accomplished by a
video-difference detector, where the information from each scene is
first converted to video signals by synchronous scanning and then sub-
tracted, leaving only the differences for display. For two video signals,
each obtained from scanning of the same geographic area and each con-
taining a target that does not appear in the corresponding scene (as
shown in Figure 18),the video-difference signal (A-B) detects not only
the existence of the changes but also their polarity. A target in video
A that does not appear in video B results in a positive-difference
signal. (For this illustration, a target is defined as a clear area in the
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fo`"Welease 2002/01/02 : CIA- -
SECTION III - RECOMMENDED SYSTEM
Figure 15 - Static Match Curve for Magnification
Figure 16 - Error Signal for Magnification
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6 -30-
t t. I_ t t t.. I t I
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
LONGITUDINAL
REFERENCE SIGNAL
PHASE
DETECTOR
LONGITUDINAL LATERAL ERROR SIGNAL
ERROR SIGNAL /^\
DEFLECTIVE
SCREEN
LONGITUDINAL
SERVO
LATERAL
SERVO ,
DEFLECTION
CIRCUITRY
2-PHASE
GENERATOR
RASTER
GENERATOR
COMPARISON
MON IT OR
SCALE FACTOR
CHANGE LENSES
\ 4 .
MAGNIFICATION
SERVO
REFERENCE
SIGNAL
GENERATOR
PHASE
DETECTOR
--- OPTICAL PATH FOR REGISTRATION
OPTICAL PATH FOR DIFFERENCE DETECTION
----- MECHANICAL MOTION
PH OTO MU LTIP LIER
CONSOLE
CONTROLS
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
SCENE
AZIMUTH
SERVO
a I I
STATINTL
STATINTL
Approved FctQelease 2002/01/02: CIA-RDP7 -
SECTION III - RECOMMENDED SYSTEM
transparency). A target in video B with none in video A results in a
negative-difference signal.
A block diagram of a simple video-difference detector is shown in
Figure 19. In this implementation the two scenes in the form of trans-
parencies are scanned with a flying-spot scanner (fss) and a split-
optical system. The video signals obtained from the two phototubes
are amplified and subtracted and used to modulate the intensity of a
monitor crt. The deflection of the monitor crt is synchronous with the
deflection of the flying-spot scanner so that the scene geometry is
maintained. The gain of the video channels is controlled so that a "no-
difference" condition results in a gray level. A change in either scene
will appear as a very light or very dark area depending on the type of
change and the scene in which it occurs.
As described previously the difference detection technique cannot dis-
criminate between changes that have occurred in the terrain and changes
caused by such things as the movement of shadows and clouds as well as
seasonal variations, which would make variations in growth appear as
changes on the read-out display,
Several methods are available to reject (filter) certain unwanted
changes from the display. One characteristic of shadows that can be
used in a shadow rejection technique is their normal appearance as a
very high density on a positive transparency, particularly under bright
sunlight conditions when shadows are the greatest problem. Figure 20
shows an aerial photograph and a trace of the transmissivity across
the line as shown. The shadowed areas have the lowest transmissivity
and are quite apparent on the densitometer traces. It is possible to
insert a clipper into the video circuits to prevent the video signal
obtained from each scene from going below a predetermined level.
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fo elease 2002/01/02 : CIA-RDP78BO47 002600010007-6
SECTION III - RECOMMENDED SYSTEM
Figure 18 - Operation of Video-Difference Detector
DEFLECTION
CIRCUITRY
RASTER
GENERATOR
J DEFLECTION
CIRCUITRY
Figure 19 - Simplified Block Diagram of Video-Difference Detector
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6 -33-
STATINTL
Approved Fcelease 2002/01/02 : CIA-R -
SECTION III - RECOMMENDED SYSTEM
Figure 20 - Transmissivity of Shadow Areas
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
STATINTL
STATINTL
Approved Foelease 2002/01/02 : CIA-RDP78B047002600010007-6
Such a level setting can be automated or it may be adjusted manually by
viewing the output display and varying the clipping threshold level until
the shadows disappear, or are at least lessened.
Clouds normally appear as a very low density on a positive transparency.
The frequency content of data within the image of a cloud is normally
very low compared to normal scene information. Rejection of scene in-
formation from clouds can be accomplished either by clipping the video
signal at a predetermined level similar to the technique used for shadow
rejection or by using a video filter to return the video to a gray level
whenever the frequency of the scene data falls below a prescribed nor-
mal level.
Seasonal and perspective changes between scenes are the most difficult
to reject automatically. There is no current method available to filter
these effects from the displays Some seasonal changes such as foliage
may be separated by conventional frequency filtering; however, such
things as perspective changes remain difficult for machine processing.
three primary areas-
1. Image registration
2,. Methods of image data comparison
3o Implementation studies
Registration tests were conducted to investigate the relationship be-
tween the accuracy of registration and such parameters as resolution,
contrast, overlap, overlap area, azimuth, and scale-factor errors.
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved For4elease 2002/01/02 : CIA-RDP78B047 002600010007-6
SECTION III - RECOMMENDED SYSTEM
The studies were performed on two pieces of laboratory equipment: a
static scene analyzer and a dynamic scene analyzer. The static an-
alyzer was used to measure the correlation function and its derivatives,
while the dynamic analyzer was used to perform actual registrations
automatically in a manner similar to that which would be used in a
change detector.
Results of the registration studies show that the accurate registration
of two images can be performed with a wide variation in photographic
resolution and contrast. The major effect of these two photographic
parameters is a variation in equipment gain that can be compensated
for with automatic gain control.. The search operation is relatively in-
dependent of the scale-factor of the photographs. Results also indicate
that an azimuth error of as much as 1. 5 deg:can be tolerated without a
search of the two photographs in azimuth during the x and y registration
Two methods of comparing image data were investigated, namely, the
video-difference and quotient-difference techniques. Both the video-
difference and quotient-difference detectors employ essentially the
same registration process. However, the read-out or change detec-
tion mode of each is completely different, The quotient-difference
detector output is the quotient of the intensities of the two scenes being
compared. The video-difference detector determines the difference of
the intensities of the two scenes being compared.
These two techniques were compared to determine:
1. Whether or not one method offers a better means
for the observer to detect image changes
2. The sensitivity of the two methods to the detec-
tion of image changes
3. Other characteristics of each method, such as
photographic processing requirements and ease
of data processing
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved FoFs9elease 2002/01/02 : CIA-RDP78B047l 002600010007-6
Samples of the video-difference and quotient-difference techniques were
obtained from a breadboard demonstration of each technique and are
shown in Figure 21. The upper-left scene is the original. Extensions
were added to the runways in the scene at the upper right. The lower-
left photograph was generated by the quotient-difference method. The
lower-right picture is a photograph of the video-difference scene as
viewed on the monitor. The resolution of the video-difference scene is
somewhat less than the quotient-difference scene, which is due mainly
to the limit imposed by the 525-line raster scan used to connect the
scenes to electrical signals. This degradation of resolution can be
eliminated by the addition of an electronic magnification or scene en-'
largement. By scanning a smaller area of the scene with the complete
raster, the scene resolution on the read-out monitor is limited only by
the system optics and film resolution.
Both methods accurately detect the changes. In fact, the video-dif-
ference method has detected some film-emulsion defects that appear as
the small bright spots on the photograph. These defects occurred
during the generation of the new scenes from the original back-lighted
plates. Both methods have essentially the same sensitivity to the
detection of photographic changes.
Since neither method has any obvious advantage over the other in its
ability to detect changes and its sensitivity to changes, the choice of
the video-difference detector was made for the following reasons.
1, The quotient-difference detector requires an
extra step of photographic processing since
both positive- and negative-transparency scenes
are needed.
2. In the quotient-difference detector, the photo-
graphic control of scene contrast adds an addi-
tional process. If the contrast of the unchanged
areas is not equal for both scenes it is extremely
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Fot elease 2002/01/02 : CIA-RDP78B0474 002600010007-6
SECTION III - RECOMMENDED SYSTEM
Figure 21 - Two Change-Detection Methods
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Forr$elease 2002/01/02 : CIA-RDP78BO47 002600010007-6
difficult to make them cancel. This problem can
be handled in the video-difference technique by
simply changing the gain of one of the difference
amplifiers following the phototubes to compen-
sate for different contrast scenes.
In the video-difference detector, scanned read-
outs from each scene are available for elec-
tronic data processing to perform such filtering
as may be required for shadow rejection, etc.
An implementation study was directed toward an investigation into the
optical requirements and a definition of the mechanical mechanism re-
quired for registration and difference detection. The optics required
for the video-difference comparison paths consist of two optical paths,
each of which image the same crt spot on the two photographic trans-
parencies. The light that passes through each transparency is collected
by condensing lenses and directed onto individual photomultiplier tubes
(see Figure 19). This provides a means of simultaneously scanning the
two scenes with a single spot. The resolution capability, flatness of
illumination, and over-all transmissivity are some of the more impor-
tant characteristics of the optics that must be considered. The optics
must also be adjustable so that the two photographs can be registered
prior to change detection. To determine the optomechanical imple-
mentation of the change detector, the following items were considered.
Maximum longitudinal and lateral displacement of
one image relative to the other
2. Longitudinal and lateral nutation of one image
relative to the other required to generate an
error signal
3. Azimuth displacement of one image relative to the
other
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Foi r Lease 2002/01/02 : CIA-RDP78B0474 002600010007-6
4. Magnification change of one image relative to
the other
5. Magnification nutation
6. Tip and tilt adjustments of one scene relative
to the other to compensate for camera sta-
bilization differences
These studies have determined that the optics for performing the nec-
essary registration and detection functions can be fabricated with
relative freedom from shading and distortions for the 70-mm case.
The use of a zoom lens for scale-factor registration demonstrated
marginal results because of the light loss through it; therefore, the
use of an axial-lens translating device appears to be a better solution
to the problem of adjusting the scale-factor.
The resolution limit imposed by the scanning raster of the video-
difference detector was eliminated by the development of an electronic
magnification implementation permitting the photo-interpreter to en-
large an area and analyze it in detail.
6. GROWTH POTENTIAL,
The development of a change detector establishes a fundamental build-
ing block for an improved photo-intelligence processing system. In
Section II it was shown how a change detector would accelerate the first
phase of photo interpretation (screening), by comparing the "take1? of
one photo mission with that of another and displaying and recording the
changes that have taken place within common areas of the photography.
In a global satellite surveillance system in continual operation, several
change detectors would play an important role in that they could be.
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Approved Fo elease 2002/01/02: CIA-RDP78B0474 002600010007-6 STATINTL
SECTION III - RECOMMENDED SYSTEM
utilized to compare the satellite imagery on a continual basis and dis-
play the changes to a command control center. Since changes per unit
area per unit time are a measure of activity, both natural and man-
made, global or political unit activity could be measured on a daily
basis, thus contributing to alert intelligence. Such a system could also
be used by a field army commander in conjunction with battlefield sur-
veillance.
One ,growth factor of a change detector is that of indexing, or plotting.
Indexing, or plotting, is defined here as the process of referencing to
a map base the areas covered by the photography of a particular recon-
naissance mission. This function, an important one to photo interpreta-
tion, could be performed in less time by utilizing certain outputs of
the change detector,
If it is assumed that the reference photography has been plotted on a
map base the error signal generated in the registration phase of change
detector operation can then be utilized to determine the center separa-
tion of the input scene with respect to the reference scene. Since
registration occurs on a frame-by-frame basis, the separation dis-
tance is also on a frame-by-frame basis. When this information is
combined with other available information such as scale-factor, ori-
entation, and exposure numbers, the input photography can be quickly
plotted with respect to the reference photography without a direct map-
to-photo comparison. Although this technique is within the state-of-
art at this time, it has not been considered as a part of the herein pro-
posed change detector implementation.
Although the amount of automatic decisioning that appears practical at
this time is very limited, it is possible to develop an automatic read-
out detector that will locate certain types of changes. Although any
method that is now developed will be limited to the very obvious
changes, it may be helpful for certain screening operations that are
now very time consuming and hence costly when performed by humans.
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
A program is recommended to design and fabricate a change detector with
the characteristics listed in Table I.
TABLE I - CHANGE-DETECTOR DESIGN GOALS
ILLEGIB
Film input
Image registration
Position (x and y)
Orientation (0)
Scale factor (V)
Scene magnification
Film to display
Area enlargement
Output
Position read-out accuracy with
respect to reference frame
Automatic (probe) or
Manual (reticle)
Operation time
Initial setup
Subsequent
Scene resolution
70-mm roll film, 100 ft maximum
?50 percent of full scale
?90 deg for reference and input films
2 X (Minimum)
5 X
40 X
Photo-reading of 14-in, tv display
?5 percent of full scale
?5 percent of full scale
30 sec per frame (worst case)
7V
40 lines per millimeter t maximum
area enlargement
-42-
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
MW
STATINTL
Approved Fo*J, elease 2002/01/02 : CIA-RDP78B0474JA002600010007-6
It will be the primary effort of this program to develop a workable model
change detector. This change detector will incorporate data processing
techniques to reject certain unwanted data, which would normally appear
as changes. These include such items as shadows and clouds. Specific
attention will be given to the development of a console design that will
give the photo-interpreter flexibility of operation with data-handling speed.
a. Components
The equipment will consist basically of (1) a comparator unit and
(2) a display unit (see Figure 22). The comparator unit will contain
the components for performing the functions of automatically register-
ing and comparing the image data. The display unit will provide the
components for displaying the scenes being compared as well as the
output comparison data and controls required to operate the console.
The various subassemblies that will comprise the comparator and
display units are described in Table II, A detail description of each
subassembly is given under the description of Task III in Item 3, d,
of this section.
b. Operational Sequence
The operation of the equipment will be centered around a console con-
taining two tv monitors called the "comparison scene" monitor and
the "change displayb " Figure 22 is a perspective view of the console
assembly with a tentative layout of monitors and controls. The
equipment will be designed to provide readout of positions of change
and frame-number information to an electric typewriter or card
punch. The operational sequence is listed below with corresponding
scene information on each monitor:
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fo elease 2002/01/02 : CIA-RDP78B047- 4002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
Figure 2Z - Console Design
-44-
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
.r Approved Fort eelease 2002/01/02 : CIA-RDP78B0474002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
1. Insert film
Z. Identification of overlapping frames
3. Initial alignment (two minutes)
a. Azimuth
b. X and Y
c. Scale factor
d. Miscellaneous (tilt, focus, etc.
4.
Automatic registration (30 sec)
a.
b.
c.
X and Y search
Nutation
Azimuth trim
5.
Detection of significant changes
a. Comparison readout
b. Noise rejection
c. Area enlargement
d. Selection of changes
6. Position readout (with respect to frame)
a. Position computation
b. Storage
Subassembly
Comparator Unit
Film-transport
mechanism
Description
Uses 70-mm roll film, 100-foot lengths;
manual, automatic variable-speed film
advance, or both; on a frame-by-frame,
partial-frame basis, or both; frame
identification capable of ?90 deg rotation
in each channel; includes condensing op-
tics and photomultiplier read-out sensors;
tip and tilt adjustment
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
'TABLE II - COMPARATOR AND DISPLAY UNIT
Subassembly I Description
Lens displacement
Lens system capable of translation in x,
mechanism
y, and magnification; consists of lenses
and electromechanical servos and drives
in x, y, and magnification
Light source and
Backlight, lenses, and mirrors neces-
associated optics
sary to image one scene on the other;
capable of automatic transition from re-
gistration to readout
Comparator electronics
Circuitry required to drive electromech-
anical components, i. e. , servomotors in
lens-displacement mechanism, match-
point detectors, reference signal genera-
tors, etc. ; generally, electronics required
for registration; electronics required for
generation of flying-spot-scanner raster;
amplification of video signals for each
channel; high-voltage power supplies for
phototubes and crt
Display Unit
Monitors
Comparison scene monitor, change dis-
play monitor, (Both are 14-in. tv-type
monitors)
Operation controls
Frame adjust; frame advance; alignment
controls in x, y directions and scale fac-
tor for each roll of film; initiating controls
for automatic registration; noise rejection;
area enlargement; flicker option; video gain
control for contrast control (for each chan-
nel); tip and tilt control
Position read-out
?5 percent maximum error design goal,
assembly
measured at the film plane; contractor
to explore possibility of area measure-
ment; will provide analog outputs in x
and y coordinates and will display numer-
ical coordinates with respect to film
Display unit electronics
Noise rejection circuitry; all power sup-
plies, except for high voltage supplies
for crt
46 -
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6 -46-
Approved For%gelease 2002/01/02 : CIA-RDP78B0474 002600010007-6 STATINTL
Approved
r Approved Folease 2002/01/02 : CIA-RDP78B0474;0002600010007-6 STATINTL
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
After the film is loaded, the frames on each roll can be viewed on the
monitors until overlapping films are identified.
The initial alignment is performed manually. First, one or both
scenes as viewed on the monitors are oriented until the scenes are
aligned within ?2 deg. The X and Y positions of the overlapping areas
are then adjusted to occupy the same position on each monitor. The
scale-factor of the two scenes is adjusted from the display on viewing
monitor no. 2 in the difference read-out condition. Final tilt and
focus are also adjusted from the difference scene on monitor no. 2.
Subsequent frames can be registered automatically by initiation of the
registration sequence after the frame is indexed. Registration is ac-
complished by first an automatic search in the X and Y directions and
subsequently by nutation of the scenes to obtain final accurate regis-
tration.
To identify possible significant changes, the registered difference
scene is viewed to locate all possible changes between the scenes;
then changes of interest are selectedafter the employment of the noise-
rejection or area enlargement options available to the viewer.
The positions of significant changes are referenced to a fiducial mark
on the reference or input frame. Readout is accomplished automati-
cally'by the use of an electronic probe or reticle.
Operator Options
Various operator options will be available for selection at the control
panel. These options will provide operational flexibility, thus en-
abling the change detector to handle films under large variations of
scene illumination, contrast, resolution, etc. These options are:
1. Automatic correlation for registration
2. Manual correlation for registration
3. Automatic frame advance in both directions
4. Semiautomatic frame advance in both directions
41-
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved For&elease 2002/01/02 : CIA-RDP78B0474&(002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
5. Minor adjustments within frame with x-alignment
control
6. Change display monitor (right side)
a. Reference scene
b. Change display (black and white changes)
(1)
(2)
Change display minus noise
Changes can be shown in both or either
polarity (black and white)
7. Comparison display monitor (left side)
a. Comparison scene
b. Reference scene
c. Reference scene/comparison scene (flicker
technique)
8. Flicker technique (left-side monitor; variable-
speed flicker)
9. Contrast, brightness and gain adjustments (both
monitors)
10. Magnification
a. Minimum 5X
b. Area enlargement 6X to 40X, continuously
variable
11. Noise rejection, variable
12. Position readout (outputs to paper tape or punch
cards)
a. General
The program will be divided into five major tasks, each consisting
of several items of work as follows.
Task I - System Predesign
1. Optics
2. Scene registration
3. Breadboard modification
-48-
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
no Approved FoQelease 2002/01/02 : CIA-RDP78B0474002600010007-6 STATINTL
Task II - Display Data Processing Studies
1. Rejection techniques
2. Detection studies
Task III - Design and Fabrication
1. Optical and mechanical assemblies
2. Electronics
3. Console and tie-in equipment
Task IV - Checkout and Evaluation
1. Functional checkout
2. Resolution and registration tests
3. Performance tests
Task V - Support Engineering
1. Project coordination
2. Reporting
3. Travel
b.
Task I - System redesign
Predesign studies will be performed to detail certain design specifi-
cations that will be required to meet the design objectives. These
include the final choice of a lens system with adequate resolution and
fidelity to permit accurate scene registration. After the lens con-
figuration is established, tolerances on various mechanical and elec-
trical components can be established. The length of the optical path
for the assembly will be determined, thus establishing limits on a
console configuration to hold the optics, electronics, and display
monitor.
The method of film loading and frame advance must be chosen. Since
it is desirable to locate the position changes with respect to each
frame, a method of frame advance must be developed that will lend
itself to the method of measuring positions within the frame to an
accuracy of t5 percent of full scale.
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Approved Foi elease 2002/01/02 : CIA-RDP78B0474 - 002600010007-6 STATINTL
Requirements, if any, for various miscellaneous optics alignments,
such as tilt and focus, will be determined.
The choice of an electronic probe or reticle readout will be made
after investigation of the position errors that can be expected with
each method.
A basic method of masking the scenes for correlation during automa-
tic registration is required. The technique must adapt itself as the
amount of overlap between frames changes.
STATINTL
A manual scene registration assembly will be fabricated and added
to a breadboard change detector to evaluate the choice of the
lens systems as well as to provide an experimental setup for tests
of various methods of electronically filtering unwanted data (such as
shadows and clouds) from the output display, see Task II).
c. Task II - Display Data Processing Studies
Several methods of electronic rejection of unwanted data from the
display will be breadboarded and evaluated. Included, for example,
will be an implementation of the shadow rejection scheme discussed
in Section III, item 4, which rejects video amplitudes below a pre-
scribed level. A similar method of rejecting clouds from the dis-
play will also be built and tested.
Those changes that occur presently in pictures taken at different
times from different positions are seasonal and perspective changes.
There is no current method available to filter these effects from the
display. Some can be separated by conventional frequency filtering
for such things as foliage growth. Although no solution to this prob-
lem is foreseen, studies will be initiated to seek possible solutions.
Several methods of displaying image differences will be investigated.
The present method of displaying the video-differences results in
both bright and dark spots for scene changes. A method of display-
ing all changes as bright spots on a dark background or as dark spots
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved For eIease 2002/01/02 : CIA-RDP78B0474WO2600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
on a light background can be implemented by display of the square of
the difference of the video signal. This method may further reduce
the time required for an operator to locate changes between pictures
by requiring him to search for only one polarity of signal on the dif-
ference monitor.
An investigation of nonlinear amplification of the difference signals
may prove to separate further the scene changes from the background,
thus making the changes more apparent to the operator.
Various display options require further study before their incorpora-
tion into the design. The option of displaying each scene alternately
before and after viewing the difference image as well as the effective-
ness of various amounts of contrast control and such functions as
brightness, slicing, and clipping are still to be determined. The use
of a nutation technique to improve the focality of the change display
to the operator will be determined. This would consist of a nutation
of one scene with respect to the other at a fairly rapid rate. Changes
on the scene not being nutated would appear stationary while the back-
ground would move. The process must be repeated, the second scene
being kept stationary and the first being nutated.
All the experiments slated for investigation under this task will be
performed on the modified breadboard equipment that will be avail-
able from Task I. Since the results of this work will not appreciably
affect the optical and mechanical design of the change detector model,
these studies will be performed throughout the greater portion of the
over-all equipment design (see schedule). Whenever the results ap-
pear promising, however, the electronics design will be modified
accordingly.
d. Task III - Design and Fabrication
(1)
Description
Task III will consist of the design and fabrication of a change
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Aw STATINTL
Approved Fo Release 2002/01/02 : CIA-RDP78B0474&A002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAMI
(2)
STATINTL
detector with the characteristics listed in Section IV, item 1,
above. The design will be essentially as shown functionally in
Figure 17 and described in Section III. The final choice of a de-
sign configuration will depend upon the results of the predesign
study (Task I). Task III will be broken into three major areas
of activity: (1) the optical and electromechanical parts required
to perform the image registration and film handling, (2) the elec-
tronics required to perform the image registration, data com-
parison, and display data processing, and (3) the console and
tie-in equipment to integrate (1) and (2) into the final assembly.
Whenever possible, existing designs and equipment will be in-
corporated into this model. The display and read-out equipment
will be purchased for modification at-or subcontracted to
meet design specifications.
The equipment will consist of two basic units (1) a comparator
unit and (2) a monitor, control, and read-out unit (see Figure
22). The comparator unit will consist of a film-transport mech-
anism, lens displacement mechanism, the registration backlight,
and necessary electronics. The monitor, control and read-out
unit will contain the input scene and difference read-out monitors,
the operational controls, and the necessary electronics to calcu-
late the position of displayed changes. Figure 17 shows a detail
functional block diagram of the proposed system. Descriptions
of the various subassemblies follow.
Film-Handling Mechanism
The film-handling mechanism will position a frame of 70-mm
film in a film gate and index properly a strip or roll of the film
frame-by-frame. The film will be used as a transparency in the
system; thus, the handling mechanism must leave a clear path
for the optical components. Since in the comparison read-out
condition the light pattern from the crt is imaged on the film and
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Approved Fo,&Release 2002/01/02 : CIA-RDP78B04WA002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
(3)
STATINTL
the image of the associated transparency is imaged on the film
during registration, clearances must be maintained in front of
the film for these rays of light. The aperture geometry required
in front of the film is established by (1) the frame size on 70-mm
film, (2) scale-factor relationship between the two transparencies,
(3) lateral lens displacement, (4) overlap condition, and (5) the
lens aperture. Condensing optics will be installed behind the film
plane to gather the light passing through the transparency and
project it onto the face of a photomultiplier tube. Space is re-
quired to insert the backlight into one channel during the regis-
tration mode of operation. To accomplish this a mirror will be
inserted between the condensing lenses and the photomultiplier
tube. Space will be allotted within the film-handling mechanism
to accommodate this backlight mechanism. The position readout
requires either positioning of the principal point of the photograph
at a specific point relative to the comparator reference line or
measurement of the offset between the principal point and the ref-
erence line.
The film handling mechanism will be capable of accepting 100-ft
rolls of film on standard spools. These spools will be mounted
on either side of the photomultiplier tube with sprockets and
spindles driven with geneva-type drives so that either a single
sprocket hole or a single frame of information may be advanced.
A differential vernier will be employed on the film drive so that
the operator will have the choice of moving the film to any posi-
tion relative to the aperture. This feature will be included for
the handling of film from panoramic cameras.
Film Mask
During automatic registration the back-lighted scene must be
masked so that only the information common to both transparencies
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
STATINTL
Approved Fo lease 2002/01/02 : CIA-RDP78B0474 002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
is imaged on the other transparency. The registration can be
unbalanced by edge effects with an improperly masked scene.
The masked scene must be smaller than the scene it is being
compared with by a margin equal to the search excursion plus
nutation radius. Various amounts of overlap, azimuth orienta-
tion differences, and scale-factor differences influence the man-
ner and degree of masking required. The square format of a 70-
mm frame helps define the amount of masking required for a given
situation where the overlap, orientation, and scale-factor values
are available. The overlap can be measured by the offsets of the
lenses of the lateral and longitudinal servos, provided the frames.
are registered on the centerline of the comparator unit.
(4) Azimuth Orientation Control
The azimuth position control performs the function of orienting
one frame with respect to the other and both frames relative to
a reference on the monitor readout. Orientation will be accom-
plished either optically by dove prisms or mechanically by phy-
sical rotation of the film. The optical method of rotating the
image will lengthen the optical path considerably due to the need
for relay type optics to insert properly the dove prism. The
physical size of the prism will be large since a large optical
aperture will be required. Since the scale-factor adjustment
will be accomplished by translation of the lens and scene (see
lens displacement mechanism), centering and axial translation
of the prism and prism assembly is critical. Proper corrections
for the aberrations due to the effectively thick inclined-plane
parallel plate presents another design problem; therefore, this
method of effecting the azimuth orientation control will probably
be implemented by the mechanical rotation of the film. The film
rolls,, film transport, photomultiplier condensing optics, photo-
multiplier, and backlight assembly will be part of an integral
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Foiselease 2002/01/02 : CIA-RDP78B0474&002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
unit; therefore, if the film frame rotates, the remainder of these
items must rotate with the unit. The envisioned mechanism con-
sists of the over-all assembly of these items mounted on two
torque tube type bearings so that the axis of rotation will coin-
cide with the nadir point on the film frame. A problem area
exists due to the difference in many instances between the center
of rotation of the mechanism and the desired center of rotation
of the area of interest, namely, the center of the overlap area.
A detailed study of this problem may further require centering
of the overlap area in the center of the frame aperture rather
than positioning of the principal point to (1) the position computer,
This requires that a lateral motion be introduced into the film
position relative to the frame aperture where previously only
longitudinal motion was required.
The angular positioning of the film-handling mechanism on the
torque tube bearings will be accomplished by use of a servo
motor in closed loop operation with a synchro transmitter and
transformer for input and follow-up components. Each channel
will have an individual input from a control transmitter. To
orient both channels simultaneously two gaged synchro differ-
entials will be used, each of the two being inserted between the
transmitter and transformer of the individual channels.
(5) Lens Displacement Mechanism
The registration of one scene with respect to the other will be
accomplished by mounting the lenses, which image the crt ras-
ter on the transparencies, so that they may be translated along
two orthogonal directions for registration in an X and Y coordi-
nate system. The translation in the X and Y directions will be
accomplished with the use of servo motors so that the lens will
automatically position when the loop through the correlator is
closed.
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved FoMR+elease 2002/01/02 : CIA-RDP78B047002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
Position readouts will be mounted on the mechanism in the form
of potentiometers in such a manner that deflection from a center
position can be electrically detected. These readouts are re-
quired to supply information of the position of the optical center-
line related to the principal .point to (1) the position computer,
(2) the matchpoint storage unit, (3) the masking devices, and (4)
the position control on the console.
The deflection of the image from the center position must be
measured and the positional information is required in the mask-
ing device, the positioning control on the console, the matchpoint
storage unit, the change position computer, and the FSS deflec-
tion circuitry involved in the manual and automatic image en-
largement. The masking device and the automatic image en-
largement are dependent upon X and Y lens position, and the
mask limits must be computed so that the information will not
be under-scanned nor over-masked. The processing of the X
and Y lens position information may lead to the situation where
it may be advantageous to use a single pickoff on each of the lens
motions and follow-up servos on the computing mechanism. Since
these functions require inputs from both channels and from both
X and Y deflections of each channel the deciding factor will be
the interaction of the servos on each other.
The difference in scale-factor requires registration of one scene
relative to the other. Each channel will have a mechanism to
translate the lens and transparency axially to effect a change in
magnification in each channel. The initial point for translating
the lens and scene in each channel will be that which allows the
raster on the crt to cover fully the 2-1/4 by 2-1/4 frame of the
70-mm film. The scale-factor mechanisms will reduce the
coverage of the crt on each transparency alternately.
-56-
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved For (ease 2002/01/02 : CIA-RDP78B047 002600010007-6
(6) Comparator Unit Electronics
The comparator unit will contain the electronic circuitry asso-
ciated with the registration process. In general this circuitry
will supply the electromechanical components, such as servo
motors, required to register the two scenes automatically. The
items requiring design effort include: servo amplifiers, match-
point detectors (X and Y search), reference signal generators,
and phase detectors. In addition, the development of narrow-
band amplifiers following one of the phototubes will be required
to separate the X and Y and scale-factor error signals. The
packaging configuration of the above items will necessarily be
determined by the electromechanical components that they sup-
ply. All the electronics required to generate the raster will be
incorporated with the cathode ray tube in the comparator unit.
High-linearity yoke drivers will be designed to accept the hori-
zontal and vertical scanning wave forms generated by the syn-
chronizing generator. The yoke drivers will also be designed
to accept the variable sweep amplitudes and positional bias volt-
ages generated when the area enlargement mode is used. Much
of this design will consist of repackaging of the existing bread-
board circuitry.
To take the difference between the two scenes during the com-
parison read-out, a wide-band electronic difference amplifier
will be designed. It will accept outputs from each phototube and
amplify and supply difference video signals to the monitors. The
option of a positive or negative polarity of either scene for use
in the "flicker" comparison will also be designed into the ampli-
fier. A video band width of 10 me will be the design goal through-
out.
Breadboards of the noise-rejection circuits and other detection
methods developed during the display processing study will be
-57-
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved Fol"Celease 2002/01/02 : CIA-RDP78BO47 i(002600010007-6 STATINTL
(7)
repackaged and incorporated into the unit. Integration of these
circuits with the electronic difference amplifier will be required.
High-voltage power supplies for the cathode ray tube and photo-
tube will be purchased and installed in this unit.
Display Unit Electronics
The display unit will contain the electronic circuitry necessary
to determine the position of the change to be read-out. The type
of circuitry required will depend on the choice of read-out method
(probe or reticle) determined in the predesign studies. A simple
analog calculator will be developed to compute the amount of lens
displacement, cathode ray tube raster displacement, and film
displacement for position read-out measurements. The resultant
analog voltage outputs will be displayed on a digital-type of dis-
play or printed readout.
This unit will also contain the synchronizing generator necessary
to generate the various wave forms and pulses required by the
flying spot scanner and scene monitors, The generator will be
designed to deliver the standard television rates of 15, 750-cps
horizontal and 60-cps vertical interlaced 2 to 1. All circuitry
will be transistorized.
The low and medium voltage power supplies for the complete
change detector will be located in this unit. The tentative volt-
ages to be supplied are: ?30 v, dc, +500 v, dc, and -100 v, dc.
It is anticipated that these will be purchased commercial sup-
plies.
(8) Control Panel
A tentative layout of the control panel located below each monitor
is shown in Figure 23. Frame adjust, frame advance, and align-
ment controls will be provided for both the input scene film and
the reference film. Coarse and vernier frame position controls
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved Folease 2002/01/02: CIA-RDP78B047002600010007-6
STATINTL
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
will be provided together with controls to index each roll of film
in the forward and reverse directions manually or automatically.
Alignment controls in the X and Y directions, scale-factor, and
tilt will be provided for each roll of film. Initiating controls for
automatic registration as well as the noise rejection, area en-
largement, and "flicker" options will be provided on the control
panel below the reference and difference monitor. Equipment
on-off switches and a video gain control for each monitor will be
located as shown in Figure 23.
(9) Monitors
Two 14-in. tv-type monitors will be provided to display (1) the
new or input scene data and (2) the reference and difference read-
out data. The monitors will be modified to accept inputs from the
comparator unit and will be programmed to display data appro-
priate to the various positions within the operations sequence.
For example, during the frame advance and initial alignment of
each film roll, input and reference scenes will be displayed.
During the comparison readout the video difference image will
be displayed with either the input or reference data. The left-
hand monitor will be capable of displaying alternate registered
pictures of the input and reference scene, thus providing a flick-
er-type display at the operator's option.
e. Task IV - Checkout and Evaluation
Checkout and evaluation will be performed to ensure performance
conforming to normal standards for laboratory equipment. It is ex-
STATINTL pected that the customer will assist Min the evaluation by contri-
buting to the final design of performance tests and making recom-
mendations for design modifications that may be required to meet
the equipment design specifications.
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Fo lease 2002/01/02 : CIA-RDP7813047 002600010007-6
Y W
U U
a F z
m ^
IL
w
N
rc F ? LL Z pa
IL 3 J a
7 H 0 w ^
0
a
^ 0
a T H
g O J Q W
m O
}
WO
<
O ^W
W F O co J
w W 0 J
} Q JO O W
Q O J a O U~ ^ J
y
O
a
J
J
Q U U W O /
0 W
1I U 7 h Z N O
O w ? Z U \y
wZ
Q
.~ J
O g Ko0 d
3 Z LL
O O
W 0
W} 0
x
U
a
O
a Z
01 Z F
LL I
F ? F~
x LL
v
[C
F O
rc
O W
n
O x 3 u H
^ Z F- U U' N
a a O W Q
O
O
U
s N 0
G W a
L
III a
L
U.
L N
F N
L
0
U
a
r ZLL
O r 0 0
W
N
W
w
U
r O
Z } J
O
Z
I- 0
x
Ir
0
m
a
LL D
> LL
n I CO
Q 0 F L LL
W F J
11- O
a ?
x
U. Id
r U U
w
Q ^
O
a 00
Z
?
J
W U I-
W
W
n
Z
1 0
^ W Z
^ W U
Z Q
O 0 J CO o
W 0 4 }
J U
Q
a
o
0
f
Q o
[C
LL z
rl CLI I x Q N
F
0
a~ x 3 0
U W
a
O
U
III Q U
s W
;0
It
Z LL
L LL O O
LL
Figure 23 - Control Panel
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Approved FokRelease 2002/01/02 : CIA-RDP78B0471002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
STATINTL
The image resolution of the system will be measured. Registration
tests will be performed with standard images so that the accuracy of
registration can be determined. Representative pictures of the dis-
play readout for various types of changes will be recorded as well as
the effects of various settings of display options that are finally in-
cluded in the design.
Evaluation tests will be performed to establish the performance of
the equipment to detect various types of changes including the num-
ber of targets, positions of targets, and the sizes of targets within
the scene. Performance will be measured by the (1) time required
for an operator to locate changes and (2) the errors of omission and
commission for various types of changes. Standard frames with
various numbers and types of changes will be made for the perform-
ance tests.
Various screening times will be determined, including (1) the time
required to detect the first pertinent change within a scene, (2) the
time required to detect the remaining changes, and (3) the time re-
quired to identify the type of change.
STATINTL
f. Task V - Support Engineering
Included in the support engineering functions are (1) program adminis-
tration, (2) reporting of technical progress and program status, (3)
generation of an equipment operation and service manual, and (4) de-
livery of the change-detector model.- will coordinate the develop-
ment program in accordance with the schedule described in 4, b, this
section, and will deliver the model 10 days after completion of the
contract. Monthly progress letters, an interim progress report, a
final technical report, and an operation and maintenance manual will
be submitted as described below.
Monthly progress letters will summarize the work accomplished, ex-
plain the derivation of technical and managerial problems encountered
and actual or proposed solutions, and outline a work program for the
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
STATINTL
Approved Focelease 2002/01/02 : CIA-RDP78B0474JA002600010007-6
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
following reporting period. A statement on the indicated sufficiency
of manpower and funds for completion of the contract will be included.
The first report will be delivered within 45 days following contract
initiation, and subsequent reports will be delivered monthly there-
after throughout the contract. Three nonreproducible copies will be
provided each period,
Fifty copies of an interim progress report will be provided on or be-
fore the 30th day after completion of the first year of the contract.
The interim report will cover the work accomplished in all areas.
This report will be distributed in accordance with a distribution list
furnished by the customer. A copy of the distribution list will be
bound into each copy.
The final technical report will cover all work accomplished during
the contract and will include schematic and wiring diagrams for all
equipment or test equipment and details of techniques developed under
the contract. It will be complete within itself. A preliminary draft
will be submitted for review and approval prior to distribution, After
STATINTL receipt of approval, -will make all necessary corrections, pub-
lish the reports, and. distribute them in accordance with a distribution
list furnished by the customer. A copy of the distribution list will be
bound into each copy of the report. A maximum of 50 copies will be
provided. The preliminary draft will be submitted on or before the
30th day after conclusion of the contract work and the approved re-
port will be distributed ZO days after receipt of approval.
The operation and service manual will cover the complete operation
and maintenance of the delivered change detector and will include
schematic wiring diagrams for model evaluation and a listing and
description of any test equipment required for servicing the equip-
ment. Maintenance procedures will be outlined. Two draft copies
will be submitted for customer approval 45 days before delivery of
the model. If the approved draft is received within 3 weeks, the 10
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Approved Fo\elease 2002/01/02 : CIA-RDP78B047002600010007-6 STATINTL
SECTION IV - PROPOSED DEVELOPMENT PROGRAM
copies of the manual will be delivered concurrently with the delivery
of the change detector.
a. Project Organization
i
STATINTL
64
STATINTL
senior
scientific staff will assist the project engineer in the monitoring and
technical direction of the program. The project engineer will be re-
sponsible for the technical management of the program, including
planning, schedules, review, and submittal of all technical reports
and other data to the contractor. He will be assisted by a project ad-
ministrator who will be responsible for such items as cost control,
facilities, manpower and planning schedules, and project liaison with-
in the company.
STATINTL
STATINTL
STATINTL
b. Program Schedule
The schedule for the proposed program is given in Figure 25. The
various tasks are shown with important benchmarks and phasing in-
formation noted.
c. Personnel
Approved For Release 2002/01/02 : CIA-RDP78B04747A002600010007-6
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6
Next 9 Page(s) In Document Exempt
Approved For Release 2002/01/02 : CIA-RDP78BO4747AO02600010007-6