DEVELOPMENT OF A VARIABLE GAMMA-PRODUCT REVERSAL PROCESSING SYSTEM

• Document Type: CREST
• Collection: NGA Records (Formerly NIMA)
• Document Number (FOIA) /ESDN (CREST): CIA-RDP79B00873A000800020080-7
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 7
• Document Creation Date: December 28, 2016
• Document Release Date: October 16, 2012
• Sequence Number: 80
• Content Type: MISC

Declassified in Part - Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
` ~.
DE vyLOPr~:ENT
of A
VARIABLE GAr~u~L~-PRODUCT REVERSAL PROCESSING SYSTEM
.Summary
A technique for detecting small changes between twc photo-
graphs of a similar scene requires control of_ the gamma-product
produced in the black and white reversal processing system*. It
was desirable to have gamma-products of 0.5, 1.0 and 2:0, there-
by allowing the illumination differerce to be represented as the
square root, lir_ear or square of the origir_al scene illumination.
Variable gamma-products. afford contrel.to the experimenter for
enhancing the scene differences. A further requirement of the
processing system was maximum scale (long straight-line portion
of D log E curve) and minimum toe and shoulder.. The scale of the
negative and positive subsequently produced by the reversal sys-
tem should be of equal magnitudes to insure complete transfer of
origir_al scene information.
The processing system-used for accomplishing these black and
white reversal characteristics is not unconver_tional. The formu-
lations are designed to provide gamma-product control by time and
'A Photoaranhic Technique for Chance Detectioiz'
STAT
STAT
~ STAT
Declassified in Part - Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
 Declassified in Part - Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
.temperature processing of the negative image follo~.aed by a com-'~~
pletion developer fixing solution. Selection of the appropriate
time and temperature for the first developer in combination with
the .appropriate second. developer-fixer solution determines the
gamma-product to be obtained.
Our work has employed Kodak Plus-X 35mm film for the devel-
oper formulation studies. The first developer is a low-gamma
phenidone-hydroquinone formulation with a small amount of silver..
halide solvent included to control Layer capacity and gamma. The
film is processed in a first developer which is followed by a stop
bath and washing; after which the negative may be fixed normally
or bleached in an acidic potassium dichromate solution. After
- washing, which follows the bleach solution, the film is immersed
in a clearing bath. The clearing bath removes the soluble silver
salts and prepares the emulsion layer ..for subsequent development
and fixation. Re-exposure may be effected by room light or by an
organic fogging agent, such as t-butylamine borane, which is.com-
pounded in the clearing bath, solution.
Depending upon thegamma-product desired one of-three second
developer formulations is used and processed for one time and
temperature. Using a completion second developer-fix relieves
the necessity for strict control of processing after the first
development step has been completed. Chemical treatment effected
in the clearing bath insures maximum scale, of the same magnitude
as the negative images, in addition to controlling the base + fog
of .the positive image.
2
Declassified in Part - Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
 Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
~~ ~ .
~~ PHOTOGi,_.~~PHIC TECHi~IC?UL' FOR CHA\TGE DETECTIO\T
STAT
Summary
The pl.ctographicrrccess lends itself readily to the problem of detecting small
chances or differences in two photographic records of a scene. When two successive
photographs of the .same general scene are taken sequentially (with a finite time
interval between), it is possible to isolate small, low-contrast differences. The
tecrnique is limited to the linear portion of the material's characteristic curve, and
depends on experimental adjustment of the processing and illumination constants.
Before establishing the necessary analytical relations, it will be essential to
develop some preliminary ideas. These are the basic relations which exist between
the photographic exposure and its resultant transmittance, irrespective of transfer
=__fitnction.
Preliminary Considerations:
2) The non-linear relation. between exposure and transmittance is
2) The proportionality between exposure and transmittance with two-step
(or reversal) processing, is
T (~ = b ~E (x) ~ Yl Y2 ,
and when
YlY2 = 1 ,
the relation is linear.
3) Linear transmittance variations can be eliminated through trans-illumination
of a superimposed negative/positive transparency pair, the' necessary condition for
-which is that the gamma of the second process must be unity.
Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
 Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
4} :fin etposure can be described in terms of its !'ac" and "dc" pates:
E(x)=A+q(x)
and the resulting transmittance in these terms is
alp-Y ~1 - P qn~ ,
for glow-contrast scene. The average transmittance is then given by
T (x~ = alp ~
5) Linear transmittance variations can be eliminated by simultaneous pro-~
jection of a negative/positive transparency pair, so that their screened images are
in register. Both illumination systems can be adjusted in brightness (bl and b2), so
. that for
T~2 x, =constant,
Y2 = b11 ~Tl~~
(b2 T x ~J
2 (~
(b11_ y2 T2(x)
2 l(~ .
The technique of clianga detection can now be developed. We consider two
photographic transparencies of the same scene, taken with a finite time lapse.bet~veen
them. We.must use a positive for one of them. The transparencies are characterized
by
Tl(~ = al ~~l(~J-yl ~ T3(~ = a E x v2y3
3 ~ 2(> ~
Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
 Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
I'or this pair,. using Ejn = pj + qjn, and the low-contrast approximation
T1(~ = alpl- rl ~ T3 (~ = a3p2 Y` ~`~
We confine the small change to be detected to the "ac" portion of the- signal,
and will locate it in the second transparency. Thus,
q2 n ~ q1 n + e (x~
e (x) ?a_1(x)
The dissimilarities existing between the two transparencies by virtue of their
having been modified by different modulation transfer functions will. be assumed small,
and of no significance for these considerations.
When the transparencies are superimposed, in register,
T13(x) = CTl(x~ T3(x~~
`1 - + yl Y2 Y3 qY n + Y2 Y~ E (~ - Yl ~2 Y3 q (~ e n _ Yl Y2 Y3 q 2 x J
L `P1 p2 ~ p2 p1P2 1 Plp2 1 n
Y 1 Y2 Y3
pl p2
Since the two-step (or reversal) processing of T3(~ permits exposure adjustment
so that pl can be made equal to p2, the necessary processing conditions are
Yl = Y~ Y3
Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
 Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
. ~, ~ ~ ~
T:?:c doss-pro-duct vanishes by an order-of-magnitude argument. However, the
ou~dratic version of the original scene is still present, and the change, being small,
could easily be hidden in this. noise.
If the t~vo transparencies are now projected (separately, but simultaneously),
so that their screened images are in register, and with brightness bland b2, the
resultant image will be proportional to
T13n - b1Tl(x) ? b21~3n
When we employ the low-contrast approximation,
T13n = b Tl(~ + b2 T3n~ 1 b2 T3~ Yp/3 En + b2 T3(x~. A23 -bpyl Tl. (~] qln
C 2 ~ l
The necessary condition that the scene information vanish and leave only the change .
b~ T3 (~ = bl P1 Tl ~
y2 r3 _ yl (Tl (x) P2 (~bl~
T3(x)] [pl] b2
Now the coefficient of ~l contains only constants, two of which (bl and b2)
can be continuously adjusted in the projection step. Thus, by varying the relative
illumination strength of the two projection systems, the value of the coefficient can
be made equal to unity, and
as before, or
Yl = Y2 y3
Cbl ~ ~ T2 (~) ~ ~pl J
b2 L Tl(~ J p2
Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
 Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7
5
Then
Tl ~ n = b~ Tl n + ~-'2 Tg (~ + b~ T~ ~ ~1 e (~ = C i + Yl e (~~
p2 ~ ~2
extremely low contrast, and must be amplified through subsequent high-gamma printing,
or by image intensification techniques. The key to the change detection process is
control of the processing constants, the gammas, and the adjustment of the projection
beam projection brightnesses. There is a trade-off between processing constants
and beam brightnesses ~;~hich eases. the requirement for. extreme precision on the
gamma-product. On the other hand, it is best not to have the beams differ too much
in brightness for reasons of relative visibility. Since the photographic constants can be
easily controlled, precision on the gamma-product can be routinely obtained through
a reversal system. Then the light-balancing constitutes only a small, final correction
'Thus, the original scene illuminance distribution has beer eliminated, and
the projection image now contains only the differences. The information exists at
to compensate for the variation in average. transmittances.
Declassified in Part -Sanitized Copy Approved for Release 2012/10/16 :CIA-RDP79B00873A000800020080-7