SUGGESTED PROGRAM OUTLINE

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP79B00873A000100010093-1
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 5
• Document Creation Date: December 28, 2016
• Document Release Date: November 1, 2012
• Sequence Number: 93
• Content Type: MISC

Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1)5X1
The work done to date by
general areas:
falls into two
25X1
1) Investigation of the effective exposure hypothesis.
This work is important due to its fundamental
relationship to system analysis and to such ex-
ploitation techniques as image processing. The
use of linear system analysis techniques and the
application of inverse filtration and other methods
depend on the assumption of linearity. If this
assumption is not good, then the whole fabric of
these techniques is suspect. It is generally accep-
ted that the photographic process is linear in
terms of effective exposure as long as the spatial
frequency is very low, say a few cycles per millime-
ter. At higher frequencies, one can show that the
hypothesis cannot work at least in a qualitative
sense. It is important to determine the degree
of residual non-linearity so that a positive state-
ment can be made about the result.
In the past two contract periods,
has investigated the uniformity of the transfer
function and the degree of residual non-linearity
in terms of residual harmonic distortion. One
vital question remains. The entire concept of
linearity is important primarily because of the
idea of additivity- of stimuli. If the system is
linear, then the response of the system to the
sum of a set of stimuli is the sum of the responses
Aof the system to each of the individual stimuli.
In other words, the system responds to each stimulus
as though the others were not present; This con-
cept is essential to the use of frequency plane
analysis for any purpose. In Fourier series analy-
sis, for example, we break_up.the stimulus into its
component frequencies and analyze the response of
the system to each. If these frequencies look dif-
ferent to the system, depending on how many are
present and what their amplitudes are, we are
faced with a hopeless analytical situation.
The critical experiment, therefore, in regard to
this hypothesis is the additivity experiment. We
measure the response of the system to each of
several separate stimuli, in this case, sinusoids
in intensity. We then subject the system to the
25X1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1-5X1
I___
sum of two or more stimuli and compare the
result with that predicted by the linear theory.
2) Mensuration Experimentation.
The mensuration program attempts to define, in
quantitative terms, the error to be expected
measuring the size of objects whose images are
only a few resolution elements across. The pro-
gram involves a mensuration target in both pos-
itive? and negative form which is imaged with
suitable reduction and at appropriate resolution
levels. The size of circles, triangles and lines
is measured by both conventional microcomparator
methods and by microdensitometry. While a large
number of test frames were involved, the test
design was such that there were no replicates
of any case with respect to the photographic
variables. Since the errors found are potentially
correctible, it is important that the statistics
.of these errors be known. In other words, any
particular observation could possibly be an error.
Additional testing will provide an assessment
of the potential usefulness of the data gathered
so far. In addition, the use of existing test
materials can be exploited in another way. It
is likely that mensuration will be the greatest
benefactor of image processing. In many/ cases,
images are dagraded/in such a way/that/cpmplete
knpfaledge yiethe eifecti*trarer funCtion
is not readily ,,ttainabae. It/is very/likely,
/7
1-lowe,(Ter,/thef't even a nlodest degree of image pro-
ceing wj4'1 produce,,,' a decrease in the variance
d. meas3Zement and/therefore/Contribute to higher
confidence in the/Measurement of small objects.
Since has available a sophisticated25xl
image processing package, it is possible to treat
a limited amount of the new, replicated, data
with image manipulation methods. Measurement of
object size from the computer output can then' be
accomplished and the result compared with that
obtained from manual measurement.
One additional factor should be considered. All
the tests conducted earlier were
carried out at a single performance level, for
all practical purposes. In this series of tests,
one additional resolution level, achieved by
stopping the lens, should be provided.
-2-
25X1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1)5X1
Suggested Program Outline
I. Additivity Experiment
The test materials should be unchanged. Exposures required
are in the rante of 10, 40 and 100 cycles per millimeter.
Combinations to be used are:,
10 and 40
10 and 100/
/ 40 and 10,0
/
Modutions employed/should be low, about 0.3, for one set of
experiments, and medium, about 0.6, for another. The ratio
c)if' lower to high9r frequency modulation should by/that con-
veniently available.
The tests sh4uld be of the/fOrm A, B, A+B with the data analyed
for each iniA.t stimulus. /The tests are of/A critical natur:e'and
any inconsi(stencies must/be resolved at the outset. Analysis
will be adong the lines/used previously/i.e., spatial si.iectral
L___
analysibased on effetive exposure giCm microdensitothetric
measux;ement. Resultd will be depicte6 graphically ant' com-
pared/with those e4ected from linea' superpositioyof the
/ /
results obtained/from the individual stimuli. Adequate analysis
.pie discussion must be supplied that the results/can be inter-
preted. ?
II. Mensuration Data Verification
A selected set of photographs will be produced, following the
general methods _employed in the earlier tests. Approximately
40 negatives will be prepared at each of two aperture settings.
The exact apertures used will depend on experimental conditions
available but should be such that the bar target resolution is
noticeably (20% or more) reduced in the stopped down case. The
conditions of exposure should be
'e 2 targets x 2 exposurest?me.-o x 10 replicates =40 negatives
( )
all subjected to the same processing. It is not necessary that
all replicates be used if sequential testing shows this to be
unnecessary.
The cases selected for replicates of existing data must corres-
pond to exisitng data insofar as targets, exposure times and
processing. Since the stopped down cases are not replicates
of anything, they may be taken under the same conditions as the
replicates of earlier data, but with the aperture stopped ap-
propriately.
The manual data will be produced as before, but sufficient statistical
25X1
-3-
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RD-P79B00873A000100010093-1 5X1
?
data will be provided that means, confidence limits and sources
of variance can be identified. Final data will include a com-
parison of the mean of. new data with that produced earlier,
the variance at each measuring point and the 95% confidence
limits of the mean.
A suitab i
e minm number 6)f negatives will h;\ subjected to
/
treatm t by inge 'ration /5l a digita computer. The
resul s will ?erifiy6d by r plication using thy' frames
prepared for that pOose, a d repeated/at two r olution
levAs. T6 measurement of object si will b performed by
.scaling diirectly Oom the/computer otput, saud scaling
being performed dhough'tlimes-by different o erators that
the error contrAution/connected therewith is identified.
Upon completion of this series of experiments, results will
be compared among the variou cases and a suitable report
prepared.
Specific Requirements
The following specific questions are 'to be answered:
.1) Considering the results of this and the previous work,
is there reason not to utilize the effective exposure
concept in its usual context?
Explain the conclusion of (1) analytically, substantiating
the conclusion by reference to experimental data.
3) Is the earlier mensuration data consistent with the
results of the replicated tests?
4) Are the dispersions in results and the systematic
errors involved such that it is worthwhile to con-
sider making corrections in measured data? If so,
how does one relate the corrections to real objects?
5) Does the use of digital image processing improve the
precision and accuracy of mensuration of small objects?
If so, what are the constraints with respect to
processing methods, cost and time? Compare the results
obtained with the manual data in the case where a
comparison is possible. Compare results at the two
resolution levels.
IV. Plannincj Requirements
The following steps shall be taken upon initiation of the
program:
1) Prepare a program plan, showing the detailed steps to
be taken, the person responsible -and the time scale
allowed.
2) Identify, on the project time scale, significant
)0
A 25
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1
milestone points and specify the accomplishments to be
made at each of these points. There shall be no fewer
than four each such points for each of the two major
portions of the program. -
3) Provide an experimental design which shows,the tasks to,
be performed, the plan of attack, and sufficient statistical
analysis to defend the program.
4) If at any time results are obtained which indicate that
the approach may be faulty or that experimental data
is defective, such results shall be reported to the
contract monitor without delay. It is understood that
this instruction shall apply to items of major proportion.
25X1
Declassified in Part - Sanitized Copy Approved for Release 2012/11/01: CIA-RDP79B00873A000100010093-1