'RESEARCH ON PHOTO INTERPRETER PERFORMACE' PROJECT P&DS/NPIC
Document Type:
Collection:
Document Number (FOIA) /ESDN (CREST):
CIA-RDP67B00511R000100090003-9
Release Decision:
RIPPUB
Original Classification:
C
Document Page Count:
6
Document Creation Date:
November 17, 2016
Document Release Date:
March 15, 2000
Sequence Number:
3
Case Number:
Publication Date:
May 27, 1965
Content Type:
MF
File:
Attachment | Size |
---|---|
CIA-RDP67B00511R000100090003-9.pdf | 312.38 KB |
Body:
Approved For ease 2000/08/28: CIA-RDP67B0051 00100090003-9
27 May 196+
SUBJECT "Research on Photo Interpreter Performance"
Project P&DS/NPIC
1. Enclosed is a copy of the Project Approval Request for
the Human Factors project, which was signed by Mr.
Lundahl on 18 May 196+ and forwarded to the DDI for approval.
2. In this case, the PAR also suffices for "development
objectives" which are usually written on a proposed project: since
it is the most comprehensive statement of the project, it is pro-
vided for your reference and information.
3. Having designated domestic targets of initial interest,
we are now beginning to look through photography obtained from
ACIC to evaluate the final suitability of each target for the
overall research task. Our meetings and seminars at
(9,12 June) should give the participants an unusual opportunity
to delve into all facets of the project and to correlate one
another's part and work during these intervening weeks since
the pilot study. By then, we should be ready to arrange the
actual overflights.
Development Branch, P&DS
Approved For Release 2000/08/28 : CIA-RDP67B00511 R000100090003-9
Approved For?ease 200042MA-RbP'6o051 500100090003-9
Research and Development
Project Approval Request
I. Identification
The proposed project, designated "Research on Photo Interpreter Performance,"
would follow up a pilot study on "Photographic Image Recognition as a Function of
25X1A Ground Resolution" which was conducted by Human Factors Research,
and NPIC, to prove that information from this kind of experimental research is
meaningful and useful to the photographic intelligence process. The proposed
effort, endorsed by DD/S&T on merit of the successful pilot study, would be 25X1A
funded by NPIC would make use of NPIC facilities and experimental
subjects, and would be of joint interest to DD/S&T and NPIC.
This project is provided for in the NPIC's FY64 financial plan at the
25X1A level under two categories of Object Class 700: "Viewers and Other 25X1A
Photo Interpretation Equipment" -, and "Special Techniques and Develop-25X1A
ment Studies" The remaining funds are available as a result of
deferment of other projects for further investigation.
II. Objectives
In extending the findings of the pilot study, the proposed program would
investigate three principal and basic areas of concern. Objectives would con-
sequently be:
A. To determine the relation between P.I. performance and ground resolution
of photography.
B. To assess the effects of stereo image viewing (as opposed to monocular
viewing) as well as the effect of mixed-resolution stereo pairs on P.I.
performance.
C. To investigate the effects of color photography on P.I. performance.
Findings will supply objective measurements which will serve to aid in the
development and use of collection systems and exploitation equipment.
III. Background
In NPIC's particular intelligence effort, "man" -- the photo interpreter --
is the key element and yet he remains the most unknown factor in the total picture.
9.clod
Approved For Release 2000/0GL01JlDff00511 R000100090003-9 d`Rf.om
~U"
d.W ..ding and d.clos.ific-on
Approved For ReI?e 2000/08/28CQ i1R0000090003-9
There are proven techniques for attaching numbers to systems capability
(currently in terms of "modulation transfer function"); but there is no
means, to date, of quantitatively accounting for or predicting or enhancing
human performance as it relates to the quality and kinds of materials
available in the interpretation task. This vagueness inhibits our develop-
ment programs.
In general terms, we want first to ask ourselves what threshold of
quality is incontravertibly set by accountable human factors? What degree
of image quality is really needed for specific targets (for this will vary);
and exactly what details do we want to be able to see in various targets?
When do stereo and color, for instance, provide more information to the
human visual system?
At long last, a value could be put on the direct benefits of resolution,
stereo and color:
A. Results from the resolution study will aid in assessing the relative
intelligence worth of various collection systems which yield different
ground resolutions. This objective, in a skeletal version, formed
the basis of the pilot study, but would now be specifically correlated
with types and priorities of targets suggested by the operational
components cooperating in the study. The resolution study will
establish a benchmark from which effects of other physical factors
can be measured.
B. From an investigation of stereo, NPIC would. finally have in its
hands some relative numbers for evaluating (or justifying) stereo
as it directly contributes to identifying specific kinds of targets.
C. The study on color will include investigations of P.I. performance
with monocular color images, of stereo images in which (1) both images
are in color and (2) one image is in color and the other is in black &
white. Variations in ground resolution will be introduced for both
the monocular and. stereo images. Results will be useful in deciding
and detailing the roles that color might, with some degree of
assurance, take in collection and. exploitation.
In what is a relatively unexplored area, the first facts and findings
will be necessarily basic and. therefore important, with many side benefits.
Additional studies may be carried on simultaneously -- as they have an
effect upon the overall investigation. For example: experiments at Harvard
University's Center for Cognitive Studies suggest that image identification
of degraded images can be strongly influenced by pre-conceptions, such as
the sort of pre-conception a P.I. might unconsciously have gained in his own
specialization (of target-type). Called "response perseveration," it can
decidedly interfere with recognition. It also encourages a person's first
notion about lower-grade information to stick in his mind, obstructing other
alternatives. Findings from such research will be applied to the proposed
Approved For Release 2000/08/2 MI Ebt11 R000100090003-9
Approved For Rel?e 2000/08/28CQA*i1R0000090003-9
project and, iii this case, might be incorporated as multiple-choice testing
techniques. Further exploration and application might prescribe a scheme,
for instance, of presenting collateral information to the P.I. and of
assigning (or rotating) tasks for the most objective interpretation possible.
Another obvious offshoot is an investigation of basic aptitudes, back-
ground and personality traits which are essential to the successfully per-
forming P.I.
Other factors of contiguous or future concern are:
1. contrast and brightness range 7. scene change detection
2. granularity 8. season/terrain
3. sun altitude and azimuth 9. searching and viewing time
4. obliquity 10. viewing equipment/scale
5. infrared photography 11. collateral information
6. real color v.s. false color 12. individual. P.I. differences
IV. Technical Specifications
All findings and figures will be submitted in report form and will also
be represented in graphs. Graphs, covering all important target-types, will
plot ground resolution (from a range of 1-10 feet) against target identifi-
cation. Curves for the same target-type will contrast stereo with monoscopic
viewing and color with black & white.
With the help of operational components a survey of targets of interest
(which are available within the U.S.) was designated. Photography should
include:
1. Missiles (ICBM)
2. Electronics (antennae)
3. Military (air fields and submarine sites, with nuclear facilities)
4. Storage (military, ammo, BW/CW)
5. Industry
6. Transportation (land, sea, ports and harbors)
7. Power
For the following reasons, actual mission photography is unsuitable for
the project and special flights must be flown: (1) techniques for simulating
scale and resolution of photography require known and exact processing para-
meters so that accurate transfer function can be determined. (2) Images must
be selected by specific recognition characteristics for purposes of testing.
(3) For purposes of correlation, color photography must be identical to black &
white photography and consequently must be shot at the exact same time.
Approved For Release 2000/08/20 LMEhAt11 R000100090003-9
25X1A
25X1A
25X1A
25X1A
Approved For Relee 2000/08/28CQ WMEh1ROO00090003-9
However, as a doublecheck and also in the process of specifying sites to be
photographed, the researchers will review Air Force reconnaissance (data) test
flights. If material is suitable, it will be used, It is probably from this
source that materials will be obtained for pretests, the purpose of which is to
determine meaningful alternatives for mutiple-choice answers,
The OSA/Air Force will be responsible for providing the aircraft, cameras,
and original film-processing necessary for collecting the photography. In the
event that ground. targets are required, the sponsor will. make the necessary
arrangements.
Since simulation techniques will be used for degrading resolution, the
initial photography must be of high quality (a ground resolution of 6")
Photography must also cover a long strip of terrain at least 2 mile wide. It
must provide at least 60% overlap between frames. Targets must be covered in
both color and black & white. Further specifications have been discussed with
the proposed contractors, and there is leeway for selecting the remaining
parameters in terms of our logistical convenience.
V. Contractor and Financial Arrangements
Through joint agreement between DD/S&T and NPIC,
and were asked to submit the only proposal as
sole bidder and joint contractors. They are uniquely qualified in terms of
professional competence, personnel thoroughly familiar with our operation, and
investigative techniques specially adapted for our purposes. Moreover, the 25X1A
resolution simulation techniques were developed by and are
proprietary in nature.
will serve as principal investigator representing
representing
will rest 0
will serve as principal investigator
Ultimate responsibility for the research
25X1A
VI. Coordination
From its inception the project has been coordinated with PID/NPIC and
DD/S&T. The Office of Logistics has taken the preliminary steps necessary
for contract negotiation.
A thorough survey of research and literature on the psycho-physiological
25X1A aspects of the P.I. function was part of an earlier study performed for NPIC
25X1A by in 1960. Recently, has reviewed current research
in this field with NPIC technical personnel, and it was consequently concluded
Approved For Release 2000/08/20 M 611 R000100090003-9
Approved For Rello CONMO91MAtl R0 0090003-9
that other efforts are not applicable to NPIC's standard of materials --
that is, based upon the quality of material now being acquired. It is also
felt that (1). techniques used to date in service-sponsored studies of photo
interpretation were inadequate in experimental design and did not yield
objective criteria; and (2). that these and other studies did not ask
questions analogous to NPIC interpretation tasks,
As a result, the contractors will from necessity collaborate closely
with NPIC P.I.s and will correlate questions with the kinds of intelligence
requirements presently levied on the Center. Looking to the future,
projected requirements for image recognition will also be incorporated
into the testing, since current intelligence requirements are most often
defined to fit NPIC's present capability and not necessarily to exceed it.
As a general reference on the project, will submit a 25X1A
short summary/evaluation of current and recent research which he judges
partially relevant to this proposed undertaking -- either in terms of
techniques or findings.
VII. Security
The contract will. be SC-1 and association with the Agency will be
classified at the Confidential level.
Approved For Release 2000/08/20 hWEbU11 R000100090003-9