AUTOMATIC TARGET RECOGNITION PROGRAM

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP78B04770A001700010012-1
• Release Decision: RIPPUB
• Original Classification: S
• Document Page Count: 4
• Document Creation Date: December 28, 2016
• Document Release Date: February 7, 2005
• Sequence Number: 12
• Publication Date: January 26, 1966
• Content Type: FORM

25X1
25X1
DDI
25X1
13. REQUIREMENT/AUTHORITY
Rapidly increasing acquisition of reconnaissance materials will demand vastly
accelerated imagery exploitation processes. A program to develop automated inter-.
pretation systems to aid human interpreters, is required for EPIC.
i
26 January 1966
1. PROJECT TITLE/CODE NAME
2. SHORT PROJECT DESCRIPTION
Automatic Target Recognition
A Government sponsored long range R&D program:
Program
to develop automated target recognition devices to
aid image interpreters.
5. CLASS OF CONTRACTOR 6. TYPE OF CONTRACT
Industrial R&D; Manufacturer CPFF
7. FUNDS 8. REQUISITION NO. 9. BUDGET PROJECT NO.
FY 19 $ NP-AT-0
10. EFFECTIVE CONTRACT DATE 11. SECURITY CLASS.
FY 19 66 (Begin - end) A.A. - Secret
T. - Unclassified
FY 19 67 1 April 1966 - January 1967 W. - Secret
12. RESPONSIBLE DIRECTORATE OFFICE/PROJECT OFFICER TELEPHONE EXTENSION
14. TYPE OF WORK TO BE DONE
Applied Research
15. CATEGORIES OF EFFORT
MAJOR CATEGORY
SUB-CATEGORIES
Automatic Target Recognition
Computers
Pattern Recognition
Soanning
16. END ITEM OR SERVICES FROM THIS CONTRACT/IMPROVEMENT OVER CURRENT SYSTEM. EQUIPMENT. ETC.
Delivery of specific, detailed designs and recommendations for components
and sub-systems of the ultimate ATR system. These designs will be based on
laboratory research, breadboarding, and evaluation of several potential ATR
techniques during this first contracted phase. Implications and expected (Contd)
17. SUPPORTING OR RELATED CONTRACTS (Agency & Other)/COORDINATION
This Automatic Target Recognition Program is to be awarded as a prime contract
which will incorporate a number of sub-contracts, associate contracts, and
consultant contracts with other companies and individuals who are prominent in ATR
research and development. A strong stipulation in the prime contract is (Coned)
18. DESCRIPTION OF INTELLIGENCE REQUIREMENT AND DETAILED TECHNICAL DESCRIPTION OF PROJECT (Continue on addi-
tional page if required)
The requirement for the development of automated image recognition systems
is based upon the following predicted trends in reconnaissance and intelligence
operations: (a) the acquisition rate, hence the volume, of operational imagery
will continue to increase; (b) the information content per unit area of imagery
continues to increase; (c) there are increasing requirements from intelligence
analysts for information to be derived from imagery, both in variety and degree
of detail; (d) the types of imagery will become more complex with the use of (Contd)
19. APPROVED BY AND DATE
OFFICE
DEPUTY DIRECTOR
DDCI
FORM
I I . 64 2338
pproved Fe
: -RDP78B04770A001700010012-1
Approved For Ie e se 20051Q5120
cl ass eview by A.
RE V- o-'7] D H?9e8qe6q2 +_
A
 Approved For Release O
,O : CIA-RDP78B04770A001700010012-1
0$ !I 4 b
i
R & D CATALOG FORM Continued... NP-AT-O
16. results of the long range program (5 years) are outlined in
paragraph 18, below.
17. that all other significant ATR work in the country will be
scrutinized and evaluated for potential usefulness to this program
and to avoid duplicative effort.
18. color, infrared, radar, etc; (e) the allowable time to extract
certain types of information from imagery will decrease; (f) it is
a long and uncertain process to recruit, train, and retain the very
large number of interpreters which would be required to fully satisfy
these future conditions by entirely 'manual' techniques. It is
therefore incumbent upon the Government to develop automated image
recognition systems to aid the fully qualified interpreter in
performing some of his less complicated tasks. This, in turn, will
allow the interpreter to concentrate on the more complex interpretative
problems.
The types of exploitation tasks which are now considered to be
amenable to automation include the following: (a) screening of
large volumes of imagery to separate 'sterile' portions of the film
from those containing possible targets of interest to the interpreter;
(b) classification of targets into broad type-categories; (c) re-scanning
of large volumes of previous coverage to search for one-type of target
of known configuration; (d) change detection on multiple coverage of
targets; (e) automatic recognition and correlation of map area to image
area in scanning operations; (f) automatic inventory of similar
objects within an image field.
The feasibility of machine recognition of targets on aerial
photography has been demonstrated. It has been demonstrated by
various electro/optical/digital techniques, several of which have been
hardware implemented in crude prototype systems. Some of the more
promising ATR developmental systems are; the Tactical Target Recognizer
for the Army; the Natural Image Computer for GIMRADA; the Automatic
Target Recognition Device for RADC; and the Minos II by Stanford for
Ft. Monmouth. All of the systems developed to date satisfy some of
the required objectives, but all have serious deficiencies when
related to the ultimate operational requirements. These are:
(a) excessive processing time per unit area of film (the bandwidth
limit); (b) the resolution and dynamic ranges are too low; (c) versa-
tility not adequate for the many variables of target images;
(d) recognition accuracies need to be improved. These deficiencies
are considered to be the major problems to be solved in the proposed
ATR program. There now appears to be considerable promise that the
development of ATR systems can be realized which will overcome most
of these deficiencies.
src L.T
Approved For Release 2005/05/20 : CIA-RDP78B04770A001700010012-1
 Approved For Release 0 ji _1%20 : CIA-RDP78B04770A001700010012-1
18. The configuration of the ultimate ATR system, or even the sub-
systems, is unknown at this point. The purpose of the initial
contracted effort will be investigation of several sub-system tech-
niques, laboratory breadboarding of the most promising sub-systems
and thorough evaluation of their potential. As a result of this
initial effort, a fairly firm system configuration can then be
designed. Our knowledge at this date indicates that there probably
will be four functional sub-systems: an input sub-system; an image
feature extraction sub-system; a decision-making and memory sub-system
(computer); and an interpretation display and control console. A
conceptual diagram of an ATR system is shown below.
INPUT PARAMETER DECISION-MAKING
SUBSYSTEM FORMATION AND MEMORY
SUBSYSTEM SUBSYSTEM
MEASUREMENT
EXTRACTION
AND
PARAMETER
FORMATION
DISPLAY PARAMETER
DATA SELECTION
25X1
it 1
I DIRECT OPTICAL DISPLAY ,i OPERATOR
DIRECT ELECTRONIC DISPLAY CONTROL
AND
CONTROL: DISPLAY
SCREENING
RESCANNING
COUNTING
CHANGE DETECTION
CLASSIFICATION
PARAMETER CONTROL *
AUXILIARY
DECISION OUTPUT MEMORY
PLUS CONFIDENCE (CHIP FILE, TAPE,
INDICATOR CORE, ETC)
Of the fifteen roposals received for this program,
proposal was clearly the most promising techni-
ca y, an proposed the most definitive program o1an or anization,
and management. ould
be the prime contractor with as the leading
sub-contractor. Other proposed sub-contractors are:
F__ I
11 ese companies represent a good national
cross-section of ATR talent and experience. The first phase of
this program would be a concentrated and concerted nine-month
effort by all the companies concerned and would encompass the following
major tasks:
Approved For Release 2005/05/20 : CIA-RDP78BO477OA001700010012-1
25X1
RKI
25X1
 Approved For Relea$e2O0~/O5120 : CIA-RDP78B04770A001700010012-1
R & D CATALOG FORM Continued... 1ILD-A-'-0
18. (a) Problem definition to determine, more specifically, the
priority of tasks toward which an ATR system should be
initially designed; and to define NPIC requirements for
processing time, accuracy, resolution, types of targets, etc.
(b) Evaluation of existing techniques for their potential
application to EPIC requirements of the ultimate system
(as determined above)
(c) Study and evaluation of human factors which might be
involved in an ATR system. These include: man-machine
interface factors; types of display and control; and
P.I. confidence factors with relation to machine decisions
(d) Preliminary sub-system design and breadboard testing of
potential sub-systems
(e) Sub-system and system specifications
(f) Formulation of a detailed long range development program.
i
Approved For Release 2005/05/20 : CIA-RDP78B04770A001700010012-1