A COMPARISON OF LINE-SCAN AND PHOTOGRAPHIC IMAGES FOR TARGET IDENTIFICATION
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP79B00873A001600040023-9
Release Decision:
RIPPUB
Original Classification:
K
Document Page Count:
26
Document Creation Date:
December 28, 2016
Document Release Date:
August 30, 2012
Sequence Number:
23
Case Number:
Publication Date:
October 1, 1969
Content Type:
REPORT
File:
Attachment | Size |
---|---|
![]() | 813.2 KB |
Body:
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
a
a
a
a
e
e
e
e
e
a
a
~i
a
a
a
B
A COMPARISON OF
LINE-SCAN AND PHOTOGRAPHIC IMAGES
FOR TARGET IDENTIFICATION
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
0
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873A001600040023-9
r
E
DRAFT..
A COMPARISON OF LINE=SCAN AND
PHOTOGRAPHIC IMAGES FOR TARGET IDENTIFICATION
T 2G~/1 /11 c.C
12EI - ",r+ 72,3
E
E
October 1969
DRAFT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873A001600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
P
DRAFT
ACKNOWLEDGMENTS
n
LI
We sincerely thank the photointerpreters
valuable assistance in evaluating the
photography, obtaining the subjects, and
gathering the data.
iii
nR A" T
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
r Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
DRAFT
Page
1.7
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . 1
METHOD . . . . . . . . . . . . . . . . . . . . ... . 2
Preparation of the Images . . . . . . . . ... . 3
The Targets . . . . . . . . . . . . . ... . . 5
Subjects . . . . . . . . . . . . . . . . 7
Experimental Design . . . . . . . . . . .. . . . 7
Procedure . . . . . . . . . . . . . . 8
RESULTS . . . . . . . . . . . . . . . . . . . 9
DISCUSSION . . . . . . . . . . . . . . . ... . . . 14
APPENDIX A . . . . . . . . . . . . . . . . . . . A-1
A
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
7~ _ 6; INA Ir 1.
A COMPARISON OF LINE-SCAN AND
PHOTOGRAPHIC IMAGES FOR TARGET IDENTIFICATION
C
e
:1
C
P
C
G
There have been several studies of the effects of photo-
graphic ground resolution on the intelligence, output of
photo-interpreters (PIs) and intelligence analysts.. There
have been fewer studies of the effects. of line-scan imagery
on intelligence output. The purpose of the study reported
here was to make an initial determination of the relation'be-
tween line-scan and photographic imagery in terms of PI
target identification performance. The relations between the
two types of. imagery.. should,be`determined so that the de-
signers of 'line-scan systems can use the results of the re-
search on the effects photographic ground resolution has on
the performance of different PI and analyst tasks.
recently completed an experimental study to determine the
informative value of static line-scan images as a function of
two variables: (1) number of-scans per target.and (2)
sign-al`-to-noise ratio.
Photographs were taken of 20 models: 10 tanks and 10
miscellaneous vehicles, such as trucks and armored cars.
The photographs were transformed into line-scan,transparencies
so that there were either 16, 32, or 48 scans per target.
Gaussian noise was added'to the transparencies producing....
signal-to-noise ratios of 3, 5, 10, 20, and 30 for each'of
the three numbers of scans. There was also a noiseless image
for each of the three numbers of scans.
The subjects were 54 college students. Each subject
was assigned randomly to one of six groups, and each group
DRAFT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
IIIW ^1pr I
LJ viewed three transparencies that contained the same amount
of noise but, . a, different number of scans per target. The
targets were in different positions in the three transparen-
cies. The models were mounted on a board in front... of the,
subjects, and their task was to match each target image with
each target model, a task similar to matching a target image
with the target as portrayed in a PI key.
In the analysis, the tanks were treated as one class
of targets and the miscellaneous vehicles were treated as
another. Matching a target image with the target model was,
considered a correct "identification," and the percentage'
of correct identifications was computed for each target
class and each experimental condition. The results of the
study were reported in The
informative value of line-scan images as a. function of sig-
nal and noise characteristics .(white, Gaussian, signal-
independent noise), March,
1969. Some of the data are, presented in thee."Results" sec-
tion of this report...
The method used in the study reported here was similar
C
to that used byl I However, PIs not
only viewed. the line-scan imagery used in the-previous study,
they also viewed; conventional photographic . transparencies of
the same targets at five different ground resolutions..;..
Their task was to match the line-scan target images
and conventional photographic target images with the target
models. The percentage of correct identifications?was. com-
puted both as a function of number of scans per target and.
signal-to-noise ratio in the line-scan images, and as a.
function of photographic ground., resolution.
2
MO A C4
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
LiKAF T .
c
Preparation of the Images .
Following are descriptions of how the line-scan and
.photographic transparent images were prepared The
line-scan images used in the study reported here were the
same ones used and a more com-
plete description of. the image preparation is given in their
report.
The Line-scan images. The model targets were placed on
a uniform' background, and the position of each was random.
They were placed so that all of them extended an equal dis-
tance in the direction perpendicular to the scan direction.
This was done so that each' image would be formed by approxi-
mately the same number of scan lines.
A diffuse light source was used to simulate the illumi-
nation of the sky, and another point light source was used
to simulate the sun at 50? above the horizon.
Three different random arrangements of the models were
photographed with Kodak High Definition Aerial Film, Type
3404, in a 35mm camera. Each photograph was enlarged so
that the. subsequent conversion into line-scan images would
result in 16, 32, and 48 scans per target.
The enlarged photographs were transformed into line-
scan images using the =. Image Generator) and
associated Digitral.Tape Memory System2. The sampling and
reconstruction spots were identical and circularly symmetri-
cal; they had Gaussian intensity distributions with half-
amplitude diameters of 0.55mm. The scan spacing was 0.55mm.
1Scott, F. A line-scan image generator, Photo Sci. &
Eng.,-Vol. II, 5, 1967.
I? Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
The photographs were sampled at intervals of 0.55mm,
the half-amplitude of the spot diameter. The sampled'trans-
mittance was quantized into 12 bits (virtually analog
imagery) and recorded digitally on tape.
Noise was added to the quantized transmittance pro-
ducing an output image with the desired noise characteristics.
The noise elements were generated using a random-number-
generator subroutine on an SDS 9.30 computer.
The noise had a Gaussian transmittance distribution,
and different. standard deviations of the.distribution were
used to produce signal-to-noise ratios of 3, 5, 10, and 203
The signal-to-noise ratio was defined as the ratio of the
standard deviation of the signal to the standard.deviation
o f the noise. --- .,
C
P
Fourteen line-scan image transparencies were produced
with the characteristics shown in Table 1. Each cell in
the table represents a line-scan transparency characterized
by the column and row values. (Note that a transparency was
not made to represent 48 scans and a signal-to-noise ratio
of'20, because it would have been almost identical to the
noiseless 'imagery.)
Each transparency contained the images of the 20 model
targets. The position of the vehicles was the same in all
transparencies.in each row in the table but differed from
one row to the next.
3A signal-to-noise
not significantly
line-scan imagery.
ratio of 30 was also used in the
study. It was not used in the study
identification performance with it was
different from performance with noiseless
nRAI;T
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30 CIA-RDP79B00873AO01600040023-9
L L,R4wr ,
C
CHARACTERISTICS-OF THE LINE-SCAN TRANSPARENCIES
NUMBER OF
SIGNAL-TO-NOISE RATIO
SCANS PER
TARGET
3
5.
10
20
Noiseless
co
16
32
48
*A transparency was not made to represent this
cel)...
The photographic ima es. The photographic images used
in the study were made b from the original negative used
in making the 48 scan, line-scan images4. The method used
is described in detail in Appendix A, which is a copy of the
report submitted tol Lwith the photographs.
The ground resolution of the five photographs was de-
termined from the average of the three-bar target readings
made by three photographic scientists. The ground resolu-
The Targets
Figure 1 is a photograph of the models used as targets
in the study.
``After they were prepared,the photographs were cut and
the targets rearranged so that their position would not be
the same as their position in the 48 scan, line-scan images..
nR e VT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
I I it it \ /"% a a
. L.
L ~,,? .15. k
i t-
eyy
r"~R
el ~? l' ~ i
~o Cjs'! O
i. Russian T-54 Tank
2. U. S. Medium Sherman M4,
A4 Tank
3. German Tank IV/H
4. German Tiger (1) Tank
5. Joe Stalin III Tank
6. U. S. Medium M-60 Tank
7. British Centruion Tank
8. U. S. Medium Patton M-47
Tank
9. French Medium Tank AMX30
10. U. S. Light "Walker Bull-
dog" M-41 Tank
Figure 1 .
11. Tank Recovery Vehicle T-119
12. Tank Recovery Vehicle T-120 (1)
13. U. S. T-120 Tank Recovery
Vehicle
14. German 234/1 Armored Car
15. U. S. M-106 Mortar Carrier
16. German Half Track Rocket
Carrier
17. Half Track - Munition Carrier
18.. U. S. M-62, 5-Ton Wrecker
.Truck
19. German Sound Detector
20. U. S. La-Cross Missile XM 4 E2
The model targets.
f%D A VIr
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
~; w rip r~ i
TABLE 2
EXPERIMENTAL DESIGN
NUMBER OF
SCANS PER
L
SIGMA!-TO-NOISE RATIO
e
TARGET
__
3
5
10
20
16
G 1
G5
G3
G4
G5
32
G1
G2 I
G3
G4
G5
48
G1
G2
? G3
*
G5
.*This cell was not represented in the study.
Subjects
The subjects were 50 professional PIs. Their mean
experience was 5-.3 years. Eight of them were from IAS, 24
from IEG,.8 from SPAD, and 10 from
Nineteen of the subjects
battle (GOB) targets.
specialized in ground order-of-.
Experimental Design
Each subject was randomly assigned to one of five. groups
with the restriction that about the same number o.f GOB
specialists be in each group of 10: the 19 specialists were
distributed 4, 4, 4, 4, 3 in the five groups.
The subjects in Groups
2, 3, and 5 viewed three
line-scan transparencies and a photographic transparency.
The subjects in Group 4 viewed'two line-scan transparencies
nR a FT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
r
:11
:1
:1
:1
u ri v '
and a photographic transparency. All subjects viewed line-
scan transparencies that differed in the number of scans per,
vehicle, but had the same signal-to-noise ratio. Table 2
shows the experimental design.
Procedure
Each subject was seated at a light-table and given a
loop and a microscope with which to view the imagery. A set
of the 20 model vehicles arrayed in 4 x 5 matrix on a board
was placed in convenient.view`.of the subjects. Each model
was identified by a number ranging from 1 to 20.
The purpose of the experiment and rules for the,task
were explained to the subjects. The rules were as follows:
1. Match the images to the models in the order
they appear in the transparency proceeding
from the upper left-hand corner of the trans-.
p.arency to the lower right. Do not skip any
images; respond to them in order..
2. Match each image to a model and write the num-
ber of the model on your answer sheet..
3. Consider each match independently of all other
matches.. You may match the same model to more
than one image.
4.- Your. initial selection is considered as final;
do not'change your answer unless you get per-.
mission to do so.
5. Use any magnification you wish.
Work at your own pace and take breaks when you
wish. .
The subjects were given the appropriate transparencies
:JI
in the predetermined orders. Each subject completed the
matching task for a given transparency before the next. one
was given to him.
The average time taken to complete the task was about
one-and-one-half hours. '
nReFT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30 CIA-RDP79B00873AO01600040023-9
LF1Ziwlk.F'
r
C
:1
:1
C
C
The results of the study are shown in Table 3 and
Figure 2 for the tank targets and in Table 4 and Figure 3
for the miscellaneous vehicle targets.
Some of the results are immediately apparent. The
tanks were more difficult to identify correctly than the
miscellaneous targets in both the photographic and the
line-scan images: distinguishing among the tanks is simply
a more demanding perceptual task than distinguishing among
the miscellaneous targets.
For both classes of targets, higher per cent identi-
fication measures were obtained with larger signal-to-noise
ratios. For the tank targets, 48 scans per target resulted
in higher per cent identification measures than 32, and
32 resulted in higher measures than 16. For the miscellaneous
targets,.48 and 32 scans per target were better than 16 but
there was little difference between 48 and 32.
No effort has been made to smooth the functions shown
in Figures 2 and 3 because additional work must be done to
obtain the required reliability. Implied in that statement
are more targets, more imagery, and more experimental subjects.
Table. 5 shows a few of the equivalences in terms of
per cent correct identifications of tank targets, between
photographic ground resolution and line-scan image character-
istics. The table was prepared to illustrate how data from
studies like the one reported here can be used to determine
the. requirements and specifications of a line-scan system..
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
DRAFT
0
0
0
6
a
a
B
e,
a
a
B
e
e
0
e
0
e
TABLE 3
PER CENT CORRECT IDENTIFICATION OF TANKS
NUMBER OF
SIGNAL-TO-NOISE RATIO
SCANS PER
TARGET
3
5
10
20
16
18
20
27
32
56
32
31
38
52
60
83
48
32
43
75
*
98
*This cell was not. represented in the study.
TABLE 4
PER CENT CORRECT IDENTIFICATION
OF MISCELLANEOUS TARGETS
NUMBER OF
SIGNAL-TO-NOISE RATIO
SCANS PER
TARGET
3
5
10
20
16
43
44
50
64
83
32
65
59
85
95
100
48
74
79
87
*
98
*This cell was not represented in the study.
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79BOO873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30 : CIA-RDP79B00873AO01600040023-9
r
DRAFT
TABLE 5
EQUIVALENCES IN TERMS OF TARGET IDENTIFICATION
PERFORMANCE AMONG LINE-SCAN AND PHOTOGRAPHIC IMAGES
(Tank Targets Only)
El.
PER CENT
16
32
48
CORRECT
SCANS/
SCANS/
SCANS/
IDENTIFICATIONS
TARGET
TARGET
TARGET
50
39.5
9.5
6.0
S/N*
S/N
S/N
60
**
20
7.6
S/N
S/N
70
**
31
9.3
S'/N
S/N
80
**
42
18.0
S/N
S/N
90
**
**
32.1
S/N
98
**
**.
40.0
S/N
* Signal-to-noise ratio
**The imagery specified in these cells, of the table will
not yield the percent correct identifications indicated.
DwRhAFT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
IMAGE SPECIFICATIONS
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Iq
Next 1 Page(s) In Document Denied
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
2
DRAFT
The study reported here was the first effort designed
to determine the relation between line-scan and photographic
imagery. The usefulness of the results is limited because
the samples of targets and subjects were small. However,.
an,important first step has been made, and because it is
obvious line-scan systems are going to be developed, add-
itional steps must be made so that both collection and ex-
ploitation systems can be-properly designed.
14
DRAFT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
. I J I M F I
nReI;T
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
LL
2
5
July 19
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
o DIVA FT
A set of positive GEM photographs was generated by
photographing a negative original with. controlled degrada-
D tion in a 1:1 optical system. The degradations were achieved
v
II
0
D
II
0
v
0
u
v
u
u
Ii
by defocusing the optical system. The resulting 'GEMs fre-
quency spectra are shown in Figure 2. The frequency spectra
of the GEMS was measured by scanning edges in the-scene with
a microdensitometer and processing, the data with a form of
edge gradient analysis.. A GEM master negative was not.re-
qui.red as the original negative supplied had a gamma of 1.0.
The choice of an optical system for the generation of
the GEMs'was based on the ease this method affords for con-
trolled image degradation and by the_time and.cost limits
imposed on this program. A 1:1.;camera system was set up on
an optical bench and consisted of a 105mm Schneider Componon
lens, target holder and film back. Defocusing was achieved
by movement of the film back away from the target-holder by
a.micrometer stage adjustment. The original negative's
frequency spectra did not contain high frequency information
as is shown in Figure 1. This MTF curve was determined by
scanning several edges throughout the original negative
frame format and selecting a transfer function that was the
approximate, center of the data spread. Thus phase shifts
caused by defocusing was not a problem,'due to the nature
of the original'scene. Tri-bar resolution was used to de-.
termine the positions of the film back which gave resolutions
of approximately 1/2, 1/4, 1/.8, and 1/16 of that obtained
at the prime focal position., These positions were then used
to photograph the scenes, along with a .target array which
consisted of an edge, high and low contrast tri-bar targets.
The addition of these targets was made necessa-ry by the
large amounts of defocus used,. Measurements of edges within
the scene' could not be made for cases'of low resolution due
to. the large area required to properly measure theedges.
The density.differences of the edge and background in the
target array-were selected to nearly match the density
A-3
fP~ ~'"T
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873A001600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
differences in the scene. Listed in Table 1 are the
density values for the GEMs and the reference target array.
Density measurements were made on vehicles in the four
corners and in the center of the scene. The results of
measurements made on the test.target control edges. appear.
in Figure 2. Traces of reference edges contained in the
scenes for which it was possible to measure the edges (num-
ber 0, 1, and 2) are given in Figure 3. The image for the
remaining GEMs could not betraced_due to problem of the
scan length required. The data given in both these figures
is uncorrected for the microdensitometer optics and slit,
since this would only affect the level of the curves not
their relative scaling to each other.
The final GEM images were made on Kodak 5235 duplicat-
ing film process in D-76 at 68? F. to give a gamma of 1.0.
Exposures were adjusted so that the final scene densities
remained approximately the same throughout the entire GEM
set.
u
r~
ci
ci
u
J
DRAFT
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
? PM" !a ? ? PM" ? a D D
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
10
-r- T-
30 40 50
Spatial Frequency cy/mm
Figure 1. Selected Nominal Scene Transfer
Function for GEM Original Negative.
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
~. D D ? D O ? D F] F-1
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
20. 30 40
Frequency
(Cycles/mm)
50 60
Figure 2.' Target Edge Frequency Spectrum
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
? ? ? P - = D
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
10
I/- U J U 4U
60 70.
Frequency
(cycles/mm)
.Figure 3. GEM Frequency Spectrum
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30 : CIA-RDP79B00873AO01600040023-9
wn^r o
TABLE 1
DENSITY MEASUREMENTS ON GEM AND TARGET ARRAY
Original Target Array
High
Edge
Low
Contrast
Target
Contrast
D-Max
2.58
1.07
.49
D-Min
.04.
.04
.04
Density Difference
2.54
1.03.
.45
Target Array High Contrast
1
D-Max
1.42
D-Min
.09
Density Difference 1.33 .
Target Array Low Contrast
1
D-Max
1.34
D-Min
.78
Density Difference
Edge Target Array
...56
1
D-Max
1.34
D-Min
..37
Density Difference
.97
2
1.38
.09
11,..219
2
.1.31
.74
.57
2
1.32
.36
.96
GEM #
3
4
5
1.40
-1.36
1.40
.09
.09
.08
1.31
1.27
1.32
GEM #
3
4
5
1.35
1.32
1.34
.79
.78
.77
.56
.54
.57
GEM #
3
4
5
1.33.
1.41
1.35
.36
.36
.36
.97
1.05
.99
GEM-Array
Density of. Density of
GEM # Background Corner Vehicles
1
2
3;
,4
5
.27.
1.08
1.27
.25
1.04
1.25
.24
.1.06
1.28
.25.
1.03
1.28
.28
1.10
1.22
Density of
Center Vehicle
1.10
1.31
1.26
1.14,
1.29
1.26
1.13
1.32
1.23
1.15
1.33
1.28
1.19
1.38
1.25
Declassified in Part - Sanitized Copy Approved for Release 2012/08/30: CIA-RDP79B00873AO01600040023-9