You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/297497162

On the specification of the DRI requirements for a standard NATO target

Research · March 2016


DOI: 10.13140/RG.2.1.4833.9604

CITATIONS READS

0 4,547

1 author:

Philippe Arthur Jean Ghislain Chevalier


CHEVALIER SOLUTIONS, Belgium
14 PUBLICATIONS   2 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

LASERS View project

Dimensional exploration View project

All content following this page was uploaded by Philippe Arthur Jean Ghislain Chevalier on 08 March 2016.

The user has requested enhancement of the downloaded file.


On the specification of the DRI requirements
for a standard NATO target

P.A.J.G. Chevalier,
Chief Scientific Officer
OIP Sensor Systems, Westerring 21, B-9700 Oudenaarde, Belgium

November 26, 2015

Abstract

Johnson criteria are used in the DRI requirements of electro-optical systems. Confusion
about what number of cycles has be used for the DRI requirements is observed. Here we
show that the numbers to be used for two-dimensional standard NATO targets are: detection
(1 cy), recognition (2 cy) and identification (6 cy).

1 Johnson criteria
The Johnson criteria are specifications about the probability of target discrimination. These DRI
criteria were created by John Johnson in 1958 (Johnson, 1958). Johnson applied the ‘’Method
of Optical Image Transformation” shown in Figure 1. For actual, real targets, both the character-
istic size of the sub-element used for detection, recognition and identification, by the observer,
as well as the number of cycles across this characteristic size have to be taken into account
(Gal, Wittenstein, Weiss, & Schuberth, 2010). Table 1 shows a variation of cycles per minimum
dimension of the target observed in broadside view. It is interesting to note that the critical
dimension, that is the height of the M-48 tank, is h = 2.438 m . Johnson concluded from Table
1 that the target transformations were found to be independent of contrast and scene signal to
noise ratio as long as the contrast in the resolution chart was the same as the contrast in the
complex target (Johnson, 1958).

1
Figure 1: Method of Optical Image Transformation [Johnson1958].

Target Broadside View Detection [cy] Orientation [cy] Recognition [cy] Identification [cy]
Truck 0.90 1.25 4.5 8.0
M-48 Tank 0.75 1.2 3.5 7.0
StalinTank 0.75 1.2 3.3 6.0
CenturionTank 0.75 1.2 3.5 6.0
Half-track 1.0 1.50 4.0 5.0
Jeep 1.2 1.50 4.5 5.5
Command Car 1.2 1.5 4.3 5.5
Soldier (Standing) 1.5 1.8 3.8 8.0
105 Howitzer 1.0 1.5 4.8 6.0
Average 1.0 ± 0.25 1.4 ± 0.35 4.0 ± 0.8 6.4 ± 1.5

Table 1: Number of cycles to be resolved on a target for specific military tasks (Johnson criteria
of 1958).

Table 2 summarizes the target discrimination criteria. The second column in the table 2
gives the number of cycles across a target for an ensemble of observers to have a 50% chance of
completing the discrimination task. The technical report of Donohue (Donohue, 1991) defines

Discrimination level Cycles on target Description


Detection 1.0 ± 0.25 Object is of military significance
Orientation 1.4 ± 0.35 Object aspect
Recognition 4.0 ± 0.8 Class of object (Jeep, tank, etc.)
Identification 6.4 ± 1.5 Member of class

Table 2: Summary of Johnson criteria of 1958.

2
the identification requirement as I = 6.4 cy or I = 12.8 TVlines.
An identification N50 comparison for targets in the visible and infrared spectral bands re-
sulted respectively in I = 7.5 cy and I = 11.5 cy on the target (O’Connor, Driggers, Vollmer-
hausen, Devitt, & Olson, 2003). The 12 test targets are given in Figure 2.

Figure 2: NVESD visible target identification set.

The Johnson criteria have evolved over time (Sjaardema & Smith, 2015) and are now repre-
sented in Figure 3. The Figure 3 shows the substitution of an object by equivalent cycles in the
task of target discrimination. We obtain a new table 3 of Johnson criteria:

Figure 3: Present Johnson criteria used in TRM4 of IOSB.

3
Discrimination level Cycles on target Description
Detection 1 Object is of military significance
Recognition 3 Class of object (Jeep, tank, etc.)
Identification 6 Member of class

Table 3: Summary of actual Johnson criteria.

Target discrimination criteria under the label of Johnson criteria occur in optical systems
requirements. Confusion about the discrimination criteria values can occur as shown in table 2
of the article of Barela (Barela et al., 2012), that wrongly interprets the tables given in RTO-
AG-300-V26 (Sabatini & Richardson, 2010). The RTO-AG-300-V26 mentions that the current
industry criteria for 1D discrimination with 50% probability level is given by the Table 4 The

Discrimination Cycles across Meaning


level minimum dimension N50
Detection 1.0 An object is present
Recognition 4.0 The class to which the object belongs
Identification 8.0 The object is discerned with sufficient clarity to specify the type

Table 4: Current industry criterion for 1-D discrimination (50% probability level).

1D discrimination models the dimensions of the target rectangle such that the area of the real
target and the rectangle model are equal. The smaller model dimension is taken equal to the
real target critical dimension, usually the minimum dimension (Ratches et al., 1975). The 2D
discrimination with 50% probability level is given by the Table 5. The minimum dimension
N50 in the 2D discrimination is based on the geometric mean of the height h and width w of the

target and thus we have N50 = w × h . The Johnson criteria provide an approximate measure

Discrimination Cycles across Meaning


level minimum dimension N50
Detection 0.75 An object is present
Recognition 3.0 The class to which the object belongs
Identification 6.0 The object is discerned with sufficient clarity to specify the type

Table 5: Current industry criterion for 2-D discrimination (50% probability level).

of the 50% probability of discrimination. The empirically established relationships between the
levels of performance as a function of resolvable cycles across a target denoted as the target
transfer probability function (TTPF) are given in Table 6 (Ratches, 1976). Confusion about the

4
Probability of discrimination Detection [cy] Conservative Optimistic Identification [cy]
recognition [cy] recognition [cy]
1.00 3 12 9 24
0.95 2 8 6 16
0.80 1.5 6 4.5 12
0.50 1.0 4 3 8
0.30 0.75 3 2.25 6
0.10 0.50 2 1.5 4
0.02 0.25 1 0.75 2
0 0.00 0 0 0

Table 6: Target Transfer Probability Functions [Ratches1976] .

probability of discrimination occurs in table 9-5 (page 9-16) of the article of Sabatini (Sabatini
& Richardson, 2010) where the values from Ratches (Ratches, 1976) are wrongly interpreted.
The Table 6 gives the correct values as given by Ratches.
The default setting of the number of cycles in TRM4, following NATO STANAG 4347
(NATO, 1989), is one, three and six for the military tasks of detection, recognition and identi-
fication, respectively (Gal et al., 2010). These cycles must be perceptible on a 2.3 m × 2.3 m
target.

2 Luminance characteristics of a target


We consider a target seen against a terrestrial background. We derive the equation of the visual
range for this case as given in (Middleton, 1968) and use the SI photometric quantities (CCPR,
2015) in this derivation. The candela is the SI base unit for the photometric quantity luminous
intensity. We quote the definition of the candela (BIPM, 2015):

The candela is the luminous intensity, in a given direction, of a source that


emits monochromatic radiation of frequency 540 × 1012 hertz and that has a radiant
intensity in that direction of 1/683 watt per steradian.

The general form of the equation relating a given spectral radiometric quantity Xe,λ (λ) to its
corresponding photometric quantity Xv,x is given by (CCPR, 2015):
Z
Kcd
Xv,x = Xe,λ (λ)Vx (λ)dλ , (1)
Vx (λa ) λ

5
where the luminous efficacy denoted Kcd = 683 cd · sr · W−1 and λa = 555.017 nm is the
wavelength in standard air (dry air at 15 ◦ C and 101325 Pa, containing 0.03 % by volume of
carbon dioxide) at the frequency of 540 × 1012 Hz given in the definition of the candela and
Vx (λ) is any of the CIE spectral luminous efficiency functions where the index x indicates the
respective function. We have Vp (λ) representing the photopic function. We define luminance,
denoted by Lv,p and expressed as [cd · m−2 ], as the amount of luminous power Φv , that will be
detected by an eye, modeled by the photopic function Vp (λ), looking at the surface of the target
St from a particular angle of view θt . The luminance Lv,p is given by the equation:
d2 Φv
Lv,p = , (2)
dSt dΩt cos(θt )
where Lv,p is the luminance, d2 Φv is the luminous power leaving the area dSt into the cone
with solid angle dΩt , θt is the angle between the normal to the target surface and the specific
direction of propagation of the light.
Consider a large target of luminance factor βo seen against a background of luminance factor
βb where both are approximately the same range r. Assume that the visible surfaces of the target
and background are parallel planes. Assume that both surfaces receive the same illumination
from the sun and the sky. Let us replace the surface of the target and the background by a white
object and define Cw as the intrinsic contrast between this white object and the horizon sky
(Middleton, 1968). Then the intrinsic luminance of the target is given by:

Lv,p (0; t) = βt (Cw + 1)Lv,p (∞; h) , (3)

while the intrinsic luminance of the background is given by :

Lv,p (0; b) = βb (Cw + 1)Lv,p (∞; h) , (4)

where Lv,p (∞; h) is the luminance of the horizon sky. The apparent luminances at range R are
given by:

Lv,p (R; t) = βt (Cw + 1)Lv,p (∞; h) exp(−σR) + Lv,p (∞; h)[1 − exp(−σR)] (5)
Lv,p (R; b) = βb (Cw + 1)Lv,p (∞; h) exp(−σR) + Lv,p (∞; h)[1 − exp(−σR)] , (6)

where σ is the atmospheric attenuation expressed in [m−1 ].The first term originates from the
Beer-Lambert’s law and the second term represents the airlight (atmospheric veil)(WMO, 2014).
The contrast between the target and the background is:
Lv,p (R; t) − Lv,p (R; b) (βt − βb )(Cw + 1) exp(−σR)
= (7)
Lv,p (R; b) βb (Cw + 1) exp(−σR) + [1 − exp(−σR)]
Lv,p (0; t) − Lv,p (0; b) (βt − βb )
= = C(0) . (8)
Lv,p (0; b) βb

6
The term C(0) is called the inherent contrast and C(R) is the apparent contrast. Observe that
this definition of contrast is different from the contrast definition used by Johnson (Johnson,
1958) where:
Lv,p (R; t)
CJohnson (R) = . (9)
Lv,p (R; b)
We rewrite the apparent contrast and simplify the equation by elimination of βb (Cw + 1) using
equation (4) :

1
C(R) = C(0) 1−exp(−σR)
(10)
1+ βb (Cw +1) exp(−σR)
1
C(R) = C(0) Lv,p (∞;h)
. (11)
1+ Lv,p (0;b)
[exp(σR) − 1]

Lv,p (∞;h)
It is common to set Lv,p (0;b)
= 1 when the target is seen against the sky. This situation is
partially simulated in a MRC test bench when the background of the target bars has the same
luminance as the complete field of observation and where the target is partially opaque. For im-
age intensifier systems one uses STANAG 4348 where the sky luminance to target background
Lv,p (∞;h
ratio is set to Lv,p (0;b)
= 2 (NATO, 1988). It is important to realize that the sky-background
Lv,p (∞;h)
ratio Lv,p (0;b)
strongly affects the apparent contrast C(R) of the target and thus the range per-
formance of the system. This effect is illustrated in Figure 4.

Figure 4: Apparent contrast as function of the sky to target background luminance ratio and the
target range for an atmospheric attenuation σ = 0.2 km−1 .

7
We solve the equation (11) for σ to obtain:
 
1 Lv,p (0; b) C(0) − C(V ) 
σ = ln 1 + (12)
V Lv,p (∞; h) C(V )

The World Meteorological Organization (WMO)(WMO, 2014) defines the meteorological op-
tical range (MOR) as:

The meteorological optical range is the length of path in the atmosphere re-
quired to reduce the luminous flux in a collimated beam from an incandescent
lamp, at a colour temperature of 2700 K, to 5 per cent of its original value, the
luminous flux being evaluated by means of the photometric luminosity function of
the International Commission on Illumination.

The meteorological optical range M OR is measured using transmissometers (Vaisala, 2010).


The defintion of visual range given by the WMO (WMO, 2014) is:

Visual range (meteorological): Distance at which the contrast of a given object


with respect to its background is just equal to the contrast threshold of an observer
(WMO, 1992a).

The definition of visibility given by the WMO (WMO, 2014) is:

Visibility, meteorological visibility (by day) and meteorological visibility at


night are defined as the greatest distance at which a black object of suitable dimen-
sions (located on the ground) can be seen and recognized when observed against
the horizon sky during daylight or could be seen and recognized during the night
if the general illumination were raised to the normal daylight level (WMO, 1992a;
2010a).

2.1 Black target against the horizon sky


For a black target we have βt = 0 and thus Lv,p (0; t) = 0 resulting in C(0) = −1. As
Lv,t (0) = 0, no matter what the incident light, its visual range V is independent of the azimuth
Lv,p (0; b)
of the sun. Consider the situation where C(V ) = −0.02, C(0) = −1 and = 1 then
Lv,p (∞; h)
the equation (12) becomes:
 
1 −1 + 0.02  3.912
σ = ln 1 + = (13)
V −0.02 V

8
Assume a visual range V = 20 km then we find σ = 0.1956 km−1 . The equation (13) is
typically used in requirements where σ is calculated based on the measurement of the meteo-
rological visibility Vm . Observe that the use of the equation (13) is only valid for black targets
seen with respect to the sky horizon.

2.2 Grey target against the horizon sky


The intrinsic luminance of a grey target depends on that of its neighborhood and thus its visual
range V will vary with its position relative to the sun. Consider the situation where C(V ) =
Lv,p (0;b)
−0.02, C(0) = −0.33 and Lv,p (∞;h
= 1, corresponding to the situation that the target is seen
against the horizon sky, then we have:
1 3.912 ln(|C(0)|)
σ= ln(50|C(0)|) = + (14)
V V V
Assume that V = 20 km and |C(0)| = 0.33 then we find that the specific attenuation σ has to
be:
3.912 ln(|C(0)|) 3.912 ln(0.33)
σ= + = + = 0.1956 − 0.0554 = 0.1402 (15)
V V 20 20
where σ is expressed in [km−1 ]. Observe that the intrinsic contrast of the target |C(0)| =
6 1
requires a lower atmospheric attenuation parameter σ, with respect to the case in paragraph 2.1,
to obtain the visual range V .

2.3 White vertical target under an overcast sky and non-reflecting earth
Consider a completely overcast sky and a non-reflecting earth (ρ = 0) then one defines Lv,t (0) =
0.5Lv,h for a vertical white target. This results in C(0) = −0.5 Consider the situation where
Lv,p (0;b)
C(V ) = −0.02, C(0) = −0.5, Lv,p (∞;h)
= 1 and V = 20 km then we find that the specific
attenuation σ has to be:
3.912 ln(|C(0)|) 3.912 ln(0.5)
σ= + = + = 0.1956 − 0.0346 = 0.1610 (16)
V V 20 20
where σ is expressed in [km−1 ].

2.4 White vertical target under an overcast sky and reflecting earth
Consider a completely overcast sky, a ground reflectivity ρ 6= 0 and a zenith angle χ = 90 ◦
then one defines (Middleton, 1968):

Lv,p (0; t) = 0.5Lv,p (∞; h)(1 + ρ) (17)

9
for a vertical white target. This results in C(0) = 0.5(1 + ρ) − 1 = −0.5 + 0.5ρ. Consider the
Lv,p (0;b)
situation where C(V ) = −0.02, ρ = 0.2, C(0) = −0.4, Lv,p (∞;h)
= 1 and V = 20 km then we
find that the specific attenuation σ has to be:
3.912 ln(|C(0)|) 3.912 ln(0.4)
σ= + = + = 0.1956 − 0.0458 = 0.1498 (18)
V V 20 20
where σ is expressed in [km−1 ].

3 Color characteristics of a target


It is known that the atmosphere influences the perception of the color of distant targets in natural
scenes (Romero, Luzón-González, Nieves, & Hernández-Andrés, 2011).The change in color
of a target observed at a distance is a consequence of the interaction between light and the
different sized particles in the atmosphere, known as absorption and scattering processes. The
color saturation reduces as function of the distance R due to the effect of the airlight in the cone
of vision. It means that the CIE 1931 color gamut is reduced as shown in Figure 5 and Figure 6
(Romero et al., 2011). At very large distances the color characteristics of the target converges

Figure 5: CIE 1931 color gamut reduction as function of the target range on a clear day.

to the chromaticity of the horizon sky.

10
Figure 6: CIE 1931 color gamut reduction as function of the target range on an overcast sky.

4 Target modulation
In many specifications the test target is specified by its target modulation m(0). One has for the
target modulation the equation:
Lv,p (0,t)
− 1

Lv,p (0, t) − Lv,p (0, b) Lv,p (0,b)
C
Johnson − 1
m(0) = = = (19)

Lv,p (0, t) + Lv,p (0, b) Lv,p (0,t) + 1 CJohnson + 1

Lv,p (0,b)

In most of the specifications of electro-optical systems the customer specifies the modulation as
a percentage. One verifies that this equation (19) takes values in the interval [0 − 1]. Assume a
grey-grey modulation specification of 20% to 30% for a target then it has to be understood as a
calculation based on (19). The specified modulation m(0) has to be converted to C(0) and thus
for the case of m(0) = 20% we find a |C(0)| = 0.33 to be used in the range calculations.
To calculate the range performance of VIS-NIR sensors the TRM4 model (Gal et al., 2010)
uses a similar approach as in the STANAG 4347 (NATO, 1989). The model is based on the per-
ception of a standard 4-bar test pattern, and uses modified Johnson criteria to calculate ranges.
TRM4 uses the average modulation at optimum phase (AMOP) for assessing the spatial trans-
fer characteristics of the imager. The AMOP was developed and validated to characterize under
sampled imagers for which no system MTF is defined. The overall figure of merit for the VIS-
NIR sensor is the minimum difference signal perceived (MSDP). The MSDP can be expressed
in contrast.

11
5 Conclusion
Johnson criteria are used in the DRI requirements of electro-optical systems. We showed that
confusion about what number of cycles has to be used for the DRI requirements exists. We
recommend that DRI requirements of electro-optical systems comply with the default setting
of the number of cycles in TRM4, following NATO STANAG 4347 (NATO, 1989), that is one,
three and six for the military tasks of detection, recognition and identification, respectively.

12
6 References
Barela, J., Kastek, M., Firmanty, K., Trzaskawka, P., Dulski, R., & J, H. (2012). Deter-
mination of range parameters of observation devices. In R. R. E. David A. Huckridge
(Ed.), Electro-optical and infrared systems: Technology and applications ix (Vol. 8541).
SPIE.
BIPM. (2015). SI Brochure: The International System of Units (SI) [8th editions, 2006;
updated in 2014]. Retrieved from http://www.bipm.org/en/publications/
si-brochure/candela.html
CCPR. (2015). Mise en pratique for the definition of the candela and associated derived units
for photometric and radiometric quantities in the International System of Units (SI) (Tech.
Rep.). BIPM. Retrieved from http://www.bipm.org/en/publications/
mises-en-pratique/
Donohue, J. (1991). Introductory review of target discrimination criteria (Tech. Rep. No.
PL-TR-92-2129). Phillips Laboratory Air Force Systesm Command.
Gal, R., Wittenstein, W., Weiss, R., & Schuberth, W. (2010, AUG). TRM4.v1 Parameter
Description and Model Documentation (No. IOSB 2010/08).
Johnson, J. (1958, OCT). Analysis of image forming systems. In Image intensifer symposium.
Middleton, W. (1968). Vision Through the Atmosphere. University of Toronto Press.
NATO. (1988). Definition of nominal static range performance for image intensifier systems
(No. STANAG 4348).
NATO. (1989). Definition of nominal static range performance for thermal imaging systems
(No. STANAG 4347).
O’Connor, J. D., Driggers, R. G., Vollmerhausen, R. H., Devitt, N., & Olson, J. (2003). Fifty-
percent probability of identification (N50) comparison for targets in the visible and in-
frared spectral bands. Optical Engineering, 42(10), 3047-3052.
Ratches, J. A. (1976). Static Performance Model for Thermal Imaging Systems. Optical
Engineering, 15(6), 156525-156525-.
Ratches, J. A., Lawson, W. R., Bergemann, R. J., Cassidy, T. W., Swenson, J. M., & Obert,
L. P. (1975, APR). Night vision laboratory static performance model for thermal view-
ing systems (Research and Development Technical Report No. ECOM-7043). United
States Army Electronics Command, Fort Monmouth, N.J., USA: U.S. Army Electronics

13
Command, Night Vision Laboratory.
Romero, J., Luzón-González, R., Nieves, J. L., & Hernández-Andrés, J. (2011, Oct). Color
changes in objects in natural scenes as a function of observation distance and weather
conditions. Appl. Opt., 50(28), F112–F120. Retrieved from http://ao.osa.org/
abstract.cfm?URI=ao-50-28-F112
Sabatini, R., & Richardson, M. (2010). Airborne Laser Systems Testing and Analysis (RTO
AGARDograph 300 No. RTO-AG-300-V26). NATO.
Sjaardema, T., & Smith, C. (2015). History and Evolution of the Johnson Criteria (Tech. Rep.
No. SAND2015-6368). Sandia National Laboratories.
Vaisala. (2010). Vaisala Visibility Sensor FS11 for demending applications. Retrieved from
http://www.vaisala.com/Vaisala%20Documents/Brochures%20and%
20Datasheets/FS11-Datasheet-B210244EN-D-LOW-v5.pdf
WMO. (2014). Measurement of Visibility. Retrieved from https://www.wmo.int/
pages/prog/www/IMOP/publications/CIMO-Guide/Prelim-2014Ed/
Prelim2014Ed P-I Ch-9.pdf

14

View publication stats

You might also like