Article

Space Object and Background Polarization Models and

Detectability Analysis
Qiang Fu 1,2,*, Feng Zhao 1,2, Hongyi Chen 1,2 and Rui Zhu 1,2 and Yingchao Li 1,2

1 Institute of Space Optoelectronics Technology, Changchun University of Science and Technology,

Changchun 130022, China
2 School of Opto-Electronic Engineering, Changchun University of Science and Technology,

Changchun 130022, China

* Correspondence: [email protected]

Abstract: During the daytime, the commonly used optical intensity imaging techniques for
spacecraft detection usually lead to poor detection due to strong background noise. Since some
materials with low reflectivity in the visible band have higher reflectivity in the near-infrared band
and less sky background noise, this paper carries out a study of near-infrared polarization detection
techniques. In this paper, firstly, the parameters in the pBRDF model of common materials are
inverted and found to be improved compared with the literature, and secondly, a complex spherical
projection is proposed to represent the polarization degree and polarization angle information of
the sky background, and this projection has less aberration than the direct projection. Finally, from
the above, the polarization detection contrast range and the original intensity contrast range are
calculated, and it is found that the maximum of the former is higher than the latter in some angles.
Citation: Fu, Q.; Zhao, F.; Chen, H.;
Zhu, R.; Li, Y. Space Object and Keywords: micro polarizer imaging; polarization bidirectional reflection distribution function;
Background Polarization Models ground-based optical detection; complex spherical mapping
and Detectability Analysis. Appl. Sci.
2022, 12, 10714. https://doi.org/
10.3390/app122110714

Academic Editor: Edik U. Rafailov 1. Introduction

Received: 25 September 2022

As the ground-to-earth reconnaissance capabilities of satellites in orbit increase [1–
Accepted: 19 October 2022 5], ground-based exploration of such satellites is required. The background noise is low
Published: 22 October 2022 and the contrast is strong when detecting at night, but it is difficult to increase these
parameters during the day due to the influence of sunlight. Polarization detection has the
Publisher’s Note: MDPI stays
advantage of enhancing the target to increase contrast, penetrating the haze to improve
neutral with regard to jurisdictional
detection, and assisting in the extraction of the target information from the daytime noise.
claims in published maps and
Any target will produce a distinctive polarization that is dictated by its own qualities in
institutional affiliations.
the process of reflection, scattering, and electromagnetic radiation. Space objects reflect
sunlight to a high degree of polarization. This demonstrates that the surface polarization
properties of the space object play a significant role in the target’s identification. The Salfer
Copyright: © 2022 by the authors.
Grove Observatory in Ohio, USA, installed a polarization device on the photoelectric
Licensee MDPI, Basel, Switzerland. detection telescope and used this technique to detect that the polarization degree of a
This article is an open access article satellite can reach a peak of 39% at a specific time, which was carried out in order to
distributed under the terms and complete a series of polarization detection experiments on space targets.
conditions of the Creative Commons The polarization degree is around 10% when the illumination signal is strong, but
Attribution (CC BY) license can reach 40% when the light characteristics of dawn are weak, according to a number of
(https://creativecommons.org/license studies by Sanchez, which have demonstrated how the target’s radiation characteristics
s/by/4.0/). and polarization characteristics differ significantly [6]. At the White Sand Range in 2011,
the US military tested air-to-air polarization imaging for target tracking; as a result, the
false alarm rate decreased from 0.52 to 0.01 [7]. Israel conducted polarization imaging
studies in a hazy environment in 2005 [8]. The findings of the research demonstrated that
polarization imaging might enhance picture contrast and broaden the scope of action.

Appl. Sci. 2022, 12, 10712. https://doi.org/10.3390/app122110714 www.mdpi.com/journal/applsci

Appl. Sci. 2022, 12, 10712 2 of 17

Only at night can visible light intensity be detected, and during the day it is
challenging to separate target information from noise. In order for the device to detect the
target during the day, the target information needs to be extracted from the daytime noise
by the polarization method in infrared band. This study develops a polarization
bidirectional reflection distribution function while taking the ing function into
consideration. It then validates the model’s viability using parameter inversion, obtains
the contrast between intensity imaging and polarization imaging through a model
simulation of polarization angle and polarization degree, and validates the benefits of the
polarization imaging technique for space targets.

2. Basic Theory of the Polarization Bidirectional Reflection Distribution

Function Model
2.1. Bidirectional Reflection Distribution Function Model Based on Microplanes
The polarimetric scattering characteristics of surfaces can be completely described by
the polarized bidirectional reflectance distribution function (pBRDF), which was defined
by D.S. Flynn in 1995 [9].
While the actual surface of the object is rough, the actual normal z and the micro-
surface normal do not coincide, and the size of the angle between the two lines can
generally be thought of as meeting the Gaussian random distribution. This assumption
forms the foundation of the traditional polarized bidirectional reflectance distribution
function (pBRDF).
The probability density function for micro-surface elements D (ω) can be written as:

1  tan 2 α
D(ω)  exp( ) (1)
2πσ 2 cos4 α 2σ 2

where σ is the coefficient of surface roughness of the object’s material. The higher the
value of σ , the closer the object’s surface reflection is to diffuse reflection. As seen in
Figure 1, k i is the light wave vector that is incident, kr is the light wave vector that is
reflected,  =i  r is the relative azimuthal angle of the two light wave vectors, θi and
θ are the angles of the incident light vector k i and the reflected light vector kr to the
r

normal z, respectivelyz and n are the microplane and macroscopic normals of the rough
material surface of the object with an α angle, and β is the angle between the light wave
vectors that are incident and the microplane norm, where the following geometric
relationship between α and β is satisfied.

cos  2β   cos θi cos θr  sin θi sin θr cos  (2)

cos θi  cos θr
cos α  (3)
2 cos β
Appl. Sci. 2022, 12, 10712 3 of 17

Figure 1. Angular relationship of the micro-surface model.

In order to continue to derive the geometric relationship between the angles within the
microfacet element, the following two angles are introduced: The normal of the material’s
surface actually forms a plane with the incident light, as does the normal of the microplane
element.
The incident light k i forms a plane with the normal of the actual surface of the
material, and also forms a plane with the normal of the microfacet. The included angle
between the two surfaces is ηi , reflected light k r forms ηr in the same way.
After calculation, we know that ηi , ηr , θi , θr , β satisfy the following equations:

cos(ηi )   (cos θi  cos θr ) / (2 cos β)  cos θi cos β / (sin θi sin β) (4)

cos(ηr )   (cos θi  cos θr ) / (2cos β)  cos θr cos β / (sin θr sin β) (5)

In the process of reflection on the surface of rough objects, there is also a partial
scattering of light being blocked by the cutoff-masking phenomenon, which will affect the
scattering characteristics of the reflected light. Accordingly, the expression of the specular
scattering model of the micro-surface element bidirectional reflection distribution
function taking the masking function into account is as follows:

1 exp   tan α / 2σ 
2 2
1 1
fBRDF (θi , i , θr , r , λ)  G(θi , i , θr , r ) (6)
2π 4σ 2 cos 4 α cos θr cos θi

where θi and θr are the angles of the incident and reflected light norm to the surface of
the macroscopic object, respectively, and i and r are the horizontal azimuth angles of
the incident and reflected light, respectively. G(θi , i , θr , r ) is the influence of the ing
effect [10] effect from the masking effect.
Assuming i  0 , the shadowing function can be reduced to a function G(θi , θr , r )
with respect to θ i , θr , r , and the expression is derived using the approximation formula
as follows [11]:
ωp tan θip tan θrp
1
1  σr tan βp (7)
G(θ i , θr , r ) 
(1  ωp tan θip )(1  ωp tan 2 θrp ) 
2

where the angle

θ ip , θrp , βp designates, in accordance with the following geometric
relationships, the spherical projection of the angle θ i , θr , β .
 up sin α
ωp  σ p (1  )
 sin α  υ p cos α
 sin θi  sin θr cos φr
 tan θip  tan θi
 2sin α cos β
 (8)
 tan θ  tan θ sin θ r  sin θi cos φr
 rp r
2sin α cos β

 cos θi  cos β
 tan βp 
 2sin α cos β
Appl. Sci. 2022, 12, 10712 4 of 17

σ u v
where p , σr , p and p are empirical parameters connected to the surface-related
material roughness coefficients and are often given the following empirical values [12]:
σp  0.0136, σr  0.0136, up  9.0, vp  1.0 .

2.2. Multi-Parameter Polarization Bidirectional Reflection Distribution Function Model of

Rough Material Surface
The bidirectional reflection distribution function can describe the reflection process
of light rays on the surface of anisotropic objects more comprehensively. Not only the
original intensity of the light will change, but so will its polarization state, and the polari-
zation bidirectional reflection distribution function model is:
dLr θi , i ; θr , r ; λ 
M θi , i ; θr , r ; λ   (9)
dEi θi , i ; λ 

where dEi and dLr are the Stokes vectors of incident and scattered light.
Based on this, the polarization bidirectional reflection distribution function
M θi , i ; θr , r ; λ  is created. This function is a crucial addition to the bidirectional reflec-
tion distribution function because it can reflect how the polarization state of light changes
during the reflection process on an object’s surface. The target’s polarization characteris-
tics can be determined using the polarization bidirectional reflection distribution function,
and the algorithm can inverse the substance’s fundamental physical properties.
Surface scattering and body scattering are the two main types of light scattered from
material surfaces. This study assumes that body scattering and multiple scattering of light
off rough material surfaces are both parts of the same non-polarized process and concur-
rently models and analyzes them.
The Mueller matrix [13], which may describe the scattering polarization characteris-
tics of the target material, can be used to determine the relationship between the incident
light and the scattered light Stokes vector using the polarization bi-directional reflection
distribution function. In order to distinguish the effects brought on by the two scattered
lights during the surface scattering of the material, the specular coefficient k s is intro-
duced in this paper, as well as a pBRDF model that includes the surface scattering com-
f
ponent sp of the material.

f  fdd  ks f sp (10)

The body scattering component fdd and the associated expressions, and ρd was
proposed in the Le Hors work to elucidate the impact of body scattering in reflected light
on the polarization state [14], as shown in following equations:

cos θr
fddi , j  ρd  M id, j (θi , θr ,  , η) (11)
π

1  k2  R
ρd  1  Ri  (12)
1  k2 R

rs  rp
Ri  (13)
2

M id, j (θi , θr ,  , η) is the Muller matrix of the body scattering effect, the Fresnel reflec-
tance of the light that comes from the first medium and hits the target surface is known as
Appl. Sci. 2022, 12, 10712 5 of 17

Ri ,which can be calculated from the Fresnel reflection coefficients rs and rp . k2 repre-
sents the Fresnel reflectance of the scattered light that was incident on the target surface
from the initial medium. R is the target surface’s diffuse reflectance at roughly infinite
r
thickness [15]. The expressions of rs and p are shown as follows:

( A  cos β) 2  B 2
rs  (14)
( A  cos β) 2  B 2

( A  sin β cos β) 2  B2
rp  rs (15)
( A  sin β cos β)2  B2

C D
A (16)
2

C D
B (17)
2

C  4n2 k2  D2 (18)

D  n 2  k 2  sin 2 β (19)

where n is the refractive index, and k is the absorption coefficient. N = n + ik is the complex
refractive index of the material. The refractive index n, defined as the ratio of the speed of
light in a vacuum to the speed of light in a medium, reflects the degree of deflection of
light as it travels through the medium. The extinction coefficient k is related to the expo-
nential attenuation of light waves as they pass through the medium and can be thought
of as representing the absorption of light by matter.
Based on the microsurface metatheory in Section 2.1 and considering the masking
effect existing on the actual rough material surface, the representation of the surface scat-
f
tering component sp is deduced as
 tan 2 α 
exp  2 
1  2σ   G θ , θ ,   M θ , θ ,  ; η (20)
fspi , j    i r  i, j  i r 
8πσ 2 cos 4 α cos θi cos θr

where i , j denote the i  1 th row and j  1 th column in the Mueller matrix, and the
value interval is an integer from 0 to 3;  is the relative azimuth angle   i  r ; σ
is the material surface roughness.
The Kubelka–Munk model was chosen to model the body scattering in the target
scattered light [16,17], and the multiparametric polarization bi-directional reflection dis-
tribution function can be modeled by combining the surface scattering component sp
f

expression and the body scattering component fdd expression as

 tan 2 α 
exp   2 
ks  2σ   G θ , θ ,   M θ , θ ,  ; η  cos θr ρ  M d (θ , θ ,  ; η)
fi , j   i r  i, j  i r  d i,j i r (21)
8πσ 2 cos 4 α cos θi cos θr π
Appl. Sci. 2022, 12, 10712 6 of 17

Since the circular polarization component in natural light is very small and negligi-
ble, its Stokes vector when natural light is incident can be expressed as Ei  1 0 0 , and
T

consequently the Stokes vector of the reflected light is as in Equation (22):

 fsp00 fsp01 fsp02  1   f sp00 

   
Lr   f sp10 f sp11 f sp12  0    fsp10  (22)
  0   
 f sp20 fsp21 f sp22     f sp20 
f
where Lr is the reflected irradiance Stokes vector parameter and i , j is the pBRDF
Mueller matrix element. Assuming that the radiation incident on the material surface is
unpolarized natural light, the polarization of the scattered light is known as DOP. The
expression of polarization can then be deduced as:

 tan 2 α 
exp   2 
ks  2σ   G r  r
( ks fsp0,0 )  ( ks f sp2,0 )
2 2 
16πσ 2 cos 4 α cos θi cos θr
s p  
DOP   (23)
ks fsp0 ,0  tan 2 α 
exp   2 
ks  2σ   G r  r  cos θr ρ

16πσ 2 cos 4 α cos θi cos θr

s p
π
d 
As can be seen from the Formulas (19) and (23), the polarization value of a rough
surface material is determined by five physical characteristics of the material, namely n ,
k , σ , k s , and ρd . The degree of polarization and other information about the target ma-
terial are typically directly obtained by polarization imaging. Experiments cannot be used
to directly measure the five aforementioned parameters. In order to generate the five
physical parameters needed for the model, this research will use methods to invert the
degree of polarization information obtained from experiments.

2.3. Parametric Genetic Algorithm Inversion Verification

In this paper, a genetic algorithm is selected to invert five key unknown parameters
in the model, denoted as a1 , a2 , a3 , a4 and a5 . The basic parameters and operations of
the genetic algorithm are set and selected for the needs of this paper as follows.
Step 1: Create an initial population of chromosomes encoded by genes at random.
We use universal binary coding in this paper and assume that the range of individual a
values to be inverted is  amin , amax  , so the binary number of nnn bits b is

b
2 n
 1  a  amin 
(24)
 amax  amin 
The first-generation initial population consists of all the individuals compiled using
the adopted coding at the very beginning, where the genes on the individuals in the initial
population are usually generated randomly within their range of values. In this paper, we
set the initial population size to 100 randomly generated individuals.
Step 2: Calculate the fitness function value for each chromosome, end the loop itera-
tion by satisfying the output condition, and output the optimal five covariates after de-
coding; if the output condition is not satisfied, continue to run down. In this paper, the
fitness function is chosen as the standard mean squared deviation model of the simulated
value k and the experimentally measured polarization value n , to ensure the trend of
selecting individuals during the selection process.
Appl. Sci. 2022, 12, 10712 7 of 17

 ( P  P)
θi
' 2

f (n, k, σ, ks , ρ)  (25)
P
2

θi

Step 3: The 100  N individuals with relatively good fitness functions are chosen to
form a new population. In the new population, the set crossover probability and method
operation are run In the selected population, there is a chance that two individuals are
randomly chosen to exchange the same type of inverse covariates, which is known as the
crossover operation in the genetic algorithm. In other words, two new chromosomes are
formed by swapping alleles from a randomly paired pair of chromosomes. The crossover
chance is set to 0.02, and single-point, multi-point, and uniform crossover methods can all
occur at the same time.
Step 4: The set mutation probability and method operations are run in the new pop-
ulation; the inverse covariates in the population’s individuals are all given a certain prob-
ability to revalue in the range of values to produce a new individual, which is called the
mutation operation in the genetic algorithm, i.e., the genes on the chromosome mutate to
other alleles with a certain probability to produce a new chromosome. The mutation prob-
ability is 0.02, and uniform, boundary, and basic position mutations can all occur at the
same time.
Step 5: The population is run again in step 2 after selection, crossover, and mutation.
When the output condition is met, the population genetic iteration is terminated. The out-
put condition is that the difference in fitness function between the largest individual in
the population and the average fitness function of all individuals in the population is less
3
than 10 , and the algorithm’s maximum number of inheritance N cannot exceed 100.
In order to verify the reliability of the theoretical model, the polarization measure-
ments in the literature [18] are used to invert the parameters of complex refractive index
N, specularity coefficient k s and reflectivity parameter ρd of aluminum and black lac-
quer materials, and the inversion results are obtained and compared with the reference
literature to obtain the error values.
The complex refractive index of the material is N  n  ki , where n is the refractive
index and k is the absorption coefficient, both of which are wavelength-dependent.
The inversion results of the two materials after the genetic algorithm operation are
shown in Table 1, and it can be seen that the results of the inversion of the complex refrac-
tive index of the two materials are closer to the real values than those of the reference
literature [18].

Table 1. Inversion results of complex refractive index of aluminum and black paint materials.

Truth Reference Estimated

Materials
n k n k n k
Aluminum 1.24 6.6 1.37 3.97 1.231 6.574
Black Paint 1.405 0.2289 1.46 1.32 1.429 0.2398

The inverse material parameters are substituted into the multi-parameter model of
this paper and the Torrance-Sparrow model [19] to verify the accuracy of the multi-pa-
rameter model, and the simulation curves of the reflected light polarization of the two
materials under the two models vary with the angle of the incident zenith of the light
source are obtained, and the DOP simulation curves of the materials are compared with
the experimental data. Figure 2 depicts this. The results show that the actual surface of the
material is not perfectly smooth, and there is also a scattering phenomenon of body scat-
tering in addition to specular reflection of light on the surface of the material, and the
body scattering is characterized by a deflection effect, which causes the simulation
Appl. Sci. 2022, 12, 10712 8 of 17

polarization value of the two materials to be low overall, so the simulation curve of the
multi-parameter model in this paper is more consistent with the experiment.

(a) (b)
Figure 2. DOP simulation curves of materials under different models, (a) Aluminum. (b) Black paint
materials.

3. Characteristics of the Sky Background Polarization Distribution

3.1. Establishment of Rayleigh Scattering Model for Sky Background Polarization
The sun position affects the sky background polarization distribution characteristics,
and in order to introduce the relationship between the ground-based optical system and
the sun position, thehorizontal coordinate system is used. A schematic diagram of the
horizontalcoordinate system is shown in Figure 3, which can be obtained from the vertical
projection of the celestial coordinate system in the equatorial plane. The position of the
Sun is S1 , the position of the ground-based optical system is O, the plane where the optical
system is located is XOY, and OV is the direction of the space target detected by the
ground-based optical system, which intersects with the celestial sphere at point V. The
altitude angle and azimuth angle of the Sun are θS and φs , respectively, and the altitude
angle and azimuth angle of the space target detected by the ground-based optical system
can be expressed by θ and φ , respectively. South, north, east and west directions are
replaced by S, N, E, and W respectively

W S1 N

 
H
S V

O
 Y
s
S E

Figure 3. Horizontal coordinate system diagram.

Appl. Sci. 2022, 12, 10712 9 of 17

According to the study of the Rayleigh scattering process in this paper, the scattering
plane in the geodetic coordinate system is the plane where SOV is located, the scattering
angle γ is the angle between the ground-based optical system and the sun line OS and
the ground-based optical system detection direction OV, and the polarization angle can
be expressed by χ  90  ZVS .
According to the cosine theorem of spherical triangles, the scattering angle can be
derived to satisfy the following geometric relations:

π  π  π  π 
cos γ  cos   θ  cos   θS   sin   θ  sin   θS  cos  φ  φS  (26)
2  2  2  2 

The result is:

cos γ  sin θ sin θS  cos θ cos θS cos  φ  φS  (27)

The expression can be derived from the basic definition of the polarization degree as
follows:

 1  cos 2 γ 
P  Pmax   (28)
 1  cos γ 
2

where, the Pmax maximum polarization of the sky background is the maximum, and the
presence of fully polarized light in the sky background light is 1.
Before finding the polarization angle expression, it is necessary to derive the expres-
sion R  r, θ, φ  of the sunlight vector at the space position according to Rayleigh e scat-
tering law as follows:

e  v cos χ  h sin χ (29)

where, v=   cos θ cos φ,  cos θ sin φ,sin θ  , h    sin φ, cos φ, 0  , r , θ , and φ are the
coordinate components of the target in the ground level coordinate system. The expres-
sion of the space object polarization angle information χ at the R  r, θ, φ  ground level
coordinates is derived by setting the ground level coordinates of the Sun S to  r, θS , φS  .
sin θ cos θS  cos θ sin θS cos  φ  φS 
tan χ  (30)
sin  φ  φS  sin θS

3.2. Sky Background Polarization Scattering Model Complex Spherical Projection Representation
According to the polarization degree and polarization angle formula, the polariza-
tion information at any position in the sky background in the three-dimensional horizon
coordinate system can be treated by the pseudo-color algorithm to distinguish the polari-
zation information differences in different locations. In order to make the results of this
study generalizable, the three-dimensional polarization distribution is projected into a
two-dimensional polarization distribution. Figures 4 and 5 are the simulation results of
the model under the conditions of 0° solar altitude and azimuth.
The direct vertical projection method is commonly used to convert the three-dimen-
sional polarization distribution of the simulated sky background into a two-dimensional
distribution. However, the projection method lacks a rigorous correspondence, resulting
in information loss, which cannot be reverse projected, and also makes the proportion of
the zenith area greater than the actual distribution.
Appl. Sci. 2022, 12, 10712 10 of 17

(a) (b)
Figure 4. 3–D polarization distribution characteristics of Rayleigh scattering sky background, (a)
Polarization degree (b) Polarization angle.

(a) (b)
Figure 5. Polarization distribution characteristics of 2–D vertical projection of Rayleigh scattering,
(a) Polarization degree (b) Polarization angle.

In this paper, the polarization distribution characteristics of the sky background the-
ory are expressed using the complex spherical mapping theory, which effectively solves
the problem of losing polarization information during direct vertical projection [17].
Each location on the celestial sphere can be projected to the point on the plane where
the optical system is placed, according to the theory of complex spherical mapping. At
this point, the sophisticated spherical projection-based two-dimensional polarization dis-
tribution model of the sky background can be constructed. Figure 6 depicts the complex
sphere’s particular projection method. The complex sphere Z  R establishes the
xm om y m coordinate system by using  0, 0, r  as the coordinate origin. L is the line formed

by joining any two points R  x0 , y0 , z0  and z '  0, 0, r  on the celestial sphere E. Because
the normal of complex sphere z  r is not orthogonal to the line L, the complex sphere
z  r must meet the line L at Rm  xm , ym  . According to the complex sphere mapping the-
ory, each point on the celestial sphere E can be projected onto the complex sphere z  r
using the above technique.
Appl. Sci. 2022, 12, 10712 11 of 17

Z
North celestial pole
Om Ym
Xm
Rm
Solar
S R


Y
O

L

Z'

Figure 6. Complex sphere mapping relation diagram.

In a coordinate system oxyz , a complex spherical coordinate system xm oym is repre-

sented as:

zr (31)

The equation for a straight line L is expressed as:

x y zr
  (32)
x0 y0 z0  r

When actually imaging the foundation of a space target, it is only necessary to con-
sider the impact of the sky background in hemispherical space. Therefore, it can be de-
duced from the spatial geometry that any point R  r, θ, φ  in the hemisphere space on the
celestial sphere E corresponds to the coordinates of the points Rm  xm , ym  projected onto
the complex sphere.
Since only the influence of the sky background in the hemispheric space needs to be
considered when actually performing ground-based detection imaging of space targets.
Therefore, the coordinates of the point R  r, θ, φ  corresponding to the projection of any
point Rm  xm , ym  in hemispheric space on the celestial sphere E onto the complex sphere
z  r can be deduced from the spatial geometric relations as Equation (33):

xm  iym  2r tan θ / 2  exp  iφ  (33)

In summary, a two-dimensional polarization distribution model of the sky back-

ground based on the complex spherical projection theory can be established. The compar-
ison of this model with the simulation effect of the model using direct vertical projection
is shown in Figures 5 and 7. It can be found that the model using complex spherical pro-
jection derived in this section can effectively correct the aberrations caused by the direct
vertical projection method, and the percentage of the zenith region has been reduced to a
certain extent.
Appl. Sci. 2022, 12, 10712 12 of 17

(a) (b)
Figure 7. 2–D polarization distribution characteristics of sky complex spherical projection, (a) Po-
larization degree (b) Polarization angle.

4. Imaging Contrast Simulation Analysis

Simulation work was carried out for the National Astronomical Observatory of the
Chinese Academy of Sciences Changchun Satellite Observatory (43°8′ N, 125°32′ E) on
December 16, 2020, at 11:42 a.m. under specific space target relative position conditions,
where the solar zenith angle θo and azimuth angle are 29°30′ and −04°32′, respectively.
Based on the simulation work, the analysis of the effect of ground-based polarization im-
aging of the space target in this paper will be continued with the following settings.
As shown in Figure 8, it is assumed that the polyimide film, solar panel, aluminum
plate and #54 black backplane are on the same flat plate, in the same plane with the sun
and the ground-based optical system, and the relative azimuth of the sun and the ground-
based optical system and the space target in this plane is 180°.
The sun is at 180° relative azimuth to both the space target and the ground-based
optical system. The sun is regarded to be a parallel light source, and the incident parallel
light angle θi on the surface of the space target is set to 30° in this paper, and the ground-
based polarization imaging effect of the satellite observation angle θr under different al-
titude angle conditions is analyzed under the above conditions.

Solar

o i Target

Ground-based r
optical system

Figure 8. Geometric angle diagram of the detection time of the space target on the ground.

The simulation of polarization of four materials, namely polyimide film, solar panel,
aluminum panel and #54 black backsheet, is carried out first. By substituting the above
conditions θi  30 ,   180 and the physical parameters of the materials obtained by
inversion into the pBRDF model established in this paper, the relationship of θr and their
degree of polarization can be obtained, as shown in Figure 9.
Appl. Sci. 2022, 12, 10712 13 of 17

Figure 9. The relationship of DOP and θr of four satellite surface materials.

Next, the solar zenith angle θo = 29°30′ and azimuth angle −04°32′, the space target
azimuth angle 101°28′ is substituted into the sky background Rayleigh scattering model,
and the simulation of the sky background polarization value under this condition can be
obtained as follows:
In Figure 10, B represents the Babinet neutral point, Om represents the zenith center
of the sky, and the black line represents the plane including the Sun, the ground-based
optical system, and the space target. The distribution curve of the polarization value of
the sky background of the space target with the zenith angle into the relative azimuth of
the sun is represented by the line segment over Om and the sun, from which we can ob-
tain the polarization value of the sky background of the space target at any zenith angle
at this time, and the space targets at 10° and 60° zenith angles are analyzed and discussed
in this paper.

Om

Figure 10. Simulation of the polarization value of the sky background.

After the target is irradiated by sunlight, the spherical scattering model simulation is
used, and assuming that the incident light is completely passed by polarized light radia-
tion and half of the unpolarized radiation is transmitted, the luminous flux arriving at the
pupil is deduced as:
Appl. Sci. 2022, 12, 10712 14 of 17

Is  λ 1.65R 2  ρ0  1  Ps    π    cos   sin  

s   ( photons /  m 2  s ) (34)
L2  h  c πL2

where, Ps is the polarization of the space target, ρ0 is the average reflectivity of the tar-
get in the detection band, L is the space target orbit height, Planck’s constant is h , c is
the speed of light, λ is the detection of the central band of the working band, and  is
the scattered light phase angle, set the band for 1.5~1.8 μm. When the signal detection
effectiveness is guaranteed that the SNR is 6, the minimum imaging luminous flux of
space target for ground-based optical system polarization imaging is estimated to be
 s  8.472  105 photons /  m 2  s  .
The expression of the number of electrons of the space target on the target surface of
the system detector is as Equation (35) shows, the parameters are show in Table 2.

1
Ns    s  A0  τ0  τs  τp  QB  t s  1  ε2   f s (35)
k

Table 2. The parameter, definition and value of Equation (35).

Definition
number of image elements occupied by the image spot on the de-
k 4
tector target surface
A0 system pupil area 2.544 m2
τ0 system transmittance 0.9
τs average atmospheric transmittance in the operating band range 0.7

τp average transmittance of the micropolarizer array of the polariza-

0.7
tion device in front of the image elements occupied by the target
QB average quantum efficiency of the detector for the effective signal 0.65
integration time of the detector to collect the space target infor-
ts 0.025 s
mation
ε blocking ratio existing in the structure of the card-based system 0.2
transmittance coefficient of the reflected light from the space target
fs 0.13
by the spectral filter

Similarly, the expression of the number of photoelectrons that can obtain the sky
background signal that reaches the detector target in the real-time imaging method of the
micropolar array is:

1  PB
N B   B  A0  τ0  τp  QB  t B  f B  Ap2  1  ε2   (36)
2

is the number of background-generated photons, MB is

M
where Φ B  5.6  10 / 2.512
10 B

the deep-sky background equivalent magnitude, which is 22, and tB  0.025s is the inte-
gration time for the detector to collect the spatial target information, f B  0.021 is the
transmittance coefficient of the filter for the sky background noise, AP  dpix / f  0.047 ''
'

is the instantaneous field of view of a single image element of the detector, dpix  15 μm
is the image element size of the detector and f is the system focal length.
'
Appl. Sci. 2022, 12, 10712 15 of 17

In the case, PB is the polarization of the sky background. Considering the photon
noise caused by the target and the sky background, the total noise signal expression of the
system is:

N  Ns  NB  ND  ts  k  NR2 (37)

where N D  0.183e is the dark current noise of a single pixel of the detector, ts is the


integration time of the detector, N R  60 e is the readout noise of the detector. Finally,


the detection signal-to-noise ratio expression can be derived as:

Ns
SNR  (38)
N

The system’s limit detection capability will be expressed in the concept of star mag-
nitude, and its transformation relationship is:
 λ2 1 8.48  1034 100.4 bc 
 λ1 4 4  dλ 
T0  λ
8
e (1.4410 / λT0 )  1
M  2.5log   (39)
 s 
 
 
where T0 is the stellar equivalent temperature, and bc is the bolometric correction, bc
takes 0.08, and T0 is 5900 K.
Let the target be at the orbit radius of 500 km with the same distribution of sky back-
ground in the field of view. By substituting the zenith angle and azimuth angle of the
space target (A = 10°, E = 101°28′ and A = 60°, E = 10 8′, respectively) into the complex
spherical coordinate system, the polarization value PB of the space target surface mate-
rial can be found in Figure 9, and the corresponding sky background polarization value
Ps can be obtained from Figure 10. Referring to Equation (38), we know the range of the
ultimate imaging capability of the system, the range of the target-background contrast of
the original intensity imaging under the ultimate detection conditions, and the range of
the target-background contrast in the polarization degree image obtained by the polariza-
tion imaging technique. The calculation results are shown in Table 3.

Table 3. Contrast calculation result.

Extreme Imaging Capability Raw Intensity Imaging Polarization Imaging

Polarization Value
Range Contrast Contrast
A = 10°, E =
0.32 [8.86, 9.28] [0.028, 0.042] [0.006, 1.5]
101°28′
A = 60°, E =
0.89 [8.83, 9.23] [0.029, 0.043] [0.002, 0.539]
101°28′

When the system reaches the limit of detection capability, the space target brightness
and sky background brightness difference of 3~4 magnitude, the brightness of the sky
background is generally 15.85~39.82 times the brightness of the space target, when the
signal-to-noise ratio is greater than 6, the detector can be considered corresponding, but
the effective signal under this condition is susceptible to background noise interference,
imaging contrast is extremely poor.
The polarization technology, by acquiring the polarization image of the space target
and the sky background, converts the contrast between the sky background and the space
target from a value with a large difference in magnitude to a value with the same magni-
tude at certain angles, so that the system can greatly improve the target-background
Appl. Sci. 2022, 12, 10712 16 of 17

contrast index at a certain detection angle when detecting the space target with a weak
signal-to-noise ratio.
For example, the polarization of the polyimide plate can reach about 0.48 when the
macroscopic surface normal of the polyimide plate is at an angle of 62° from the center of
the field of view of the ground-based optical system at a position of 10° zenith angle and
101°28′ azimuth angle, while the polarization of the sky background is only 0.32.
Sometimes, the polarization contrast will indeed be lower than the contrast of inten-
sity imaging due to the low brightness of the polarized image caused by the addition of a
micropolarizer array, but a higher contrast of polarized imaging can be obtained by choos-
ing the most appropriate satellite observation angle. The highest value of spatial target-
background contrast can reach 1.5, which is improved by 35.71 times compared to the
maximum value of target-background contrast of 0.042 for the original intensity imaging
technique at this condition.
At this time, the target pixel value in the polarization image acquired by the system
is larger than the value of the bright sky background, which effectively reflects and uses
the ability of polarization imaging technology to “weaken strong light” and “strengthen
weak light”, so that the weak space object signal is highlighted in the strong background
noise of the daytime sky.
In summary, the space target polarization imaging technology can greatly improve
the target-background contrast index under certain angles. The difference between space
target and sky background polarization characteristics can be used to capture space tar-
gets more efficiently in the stable sky background polarization distribution, which can
effectively solve the problem of low imaging contrast caused by strong background noise
interference during daytime, and can become an important technical component of the
all-day space target ground-based observation network.

5. Conclusions
In this paper, based on the traditional BRDF model, considering that the normal of
the rough surface and the normal of the microsurface element do not coincide, a micro-
surface element BRDF model considering the masking function is established, and a
pBRDF model is further established. Aluminum and black paint are selected, and the
physical parameters in the model are verified by genetic algorithm inversion, and it is
found that the model in this paper is closer to the real value than the reference literature
model. The astronomical coordinate system and the geodetic coordinate system are intro-
duced to study the sky background polarization distribution characteristics, and the space
object polarization angle information is derived according to the Rayleigh scattering law.
A complex spherical projection is used to project the background polarization from 3–D
to 2–D, which has less aberration. The target-background contrast values of the original
intensity and the polarization intensity are obtained by selecting the zenith angles of 10°
and 60° and their azimuth angles, respectively. The polarization method is found to be
significantly higher than the original intensity method.

Author Contributions: Data curation, H.C.; Formal analysis, H.C. and Q.F.; Funding acquisition,
Q.F. and Y.L.; Methodology, Q.F.; Project administration, Q.F.; Resources, Q.F. and F.Z.; Supervi-
sion, Q.F.; Writing—original draft, H.C.; Writing—review & editing, F.Z. and R.Z. All authors have
read and agreed to the published version of the manuscript.
Funding: This research was funded by Changchun University of Science and Technology, China.
Huilin Jiang, National Natural Science Foundation of China, grant number 61890963.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The study did not report any data.
Conflicts of Interest: The authors declare no conflict of interest.
Appl. Sci. 2022, 12, 10712 17 of 17

References
1. Jiang, H.L.; Fu, Q.; Zhang, Y.L.; Jiang, L. Discussion on integrated technology of space debris laser detection, imaging and
communication. Infrared Laser Eng. 2016, 45, 7.
2. Xu, M.; Shi, H.D.; Wang, C.; Liu, Z.; Fu, Q.; Ying Chao, L.; Dong, K.Y.; Jiang, H.L. Research on the integration technology of
multi-dimensional space target detection and laser communication. Chin. J. Lasers 2021, 48, 223–235.
3. Zhang, Y.L.; Fu, Q.; Jiang, H.L.; Wang, C.; Han, L. Parameter analysis of range imaging communication integrated optical
system. Chin. J. Lasers 2017, 44, 206–212.
4. Jiang, H.L.; Jiang, L.; Fu, Q.; Dong, K.Y. Research on space debris polarization spectral imaging detection technology. J. Deep.
Space Explor. 2015, 272–277. DOI:10.15982/j.issn.2095-7777.2015.03.014.
5. Shi, X.W.; Xu, X.J. Detector material and sensitivity analysis for space-based infrared sensors. Optoelectron. Lett. 2010, 6, 27–30.
6. Jorgensen, K.; Africano, J.L.; Stansbery, E.G.; Kervin, P.W.; Hamada, K.M.; Sydney, P.F.; Pirich, A.R.; Repak, P.L.; Idell, P.S.;
Czyzak, S.R. Determining the material type of man-made orbiting objects using low-resolution reflectance spectroscopy. Int.
Soc. Opt. Photonics 2001, 4490, 237–244.
7. Ratliff, B.M.; Lemaster, D.A.; Mack, R.T.; Villeneuve, P.V.; Weinheimer, J.J.; Middendorf, J.R. Detection and tracking of RC
model aircraft in LWIR microgrid polarimeter data. In Proceedings of the Polarization Science and Remote Sensing V, San
Diego, CA, USA, 9 September 2011; pp. 29–41.
8. Namer, E.; Shwartz, S.; Schechner, Y.Y. Skyless polarimetric calibration and visibility enhancement. Opt. Express 2009, 17, 472–
493.
9. Flynn, D.S.; Alexander, C. Polarized surface scattering expressed in terms of a bidirectional reflectance distribution function
matrix. Opt. Eng. 1995, 34, 1646–1650.
10. Blinn, J.F. Models of light reflection for computer synthesized pictures. In Proceedings of the 4th Annual Conference on
Computer Graphics and Interactive Techniques, San Jose, CA, USA, 20–22 July 1977; pp. 192–198.
11. Meng, Q.Y.; Zhang, W.; Long, F.N. Analysis on the Detection Ability of Visible Cameras for Space based Space Targets. Infrared
Laser Eng. 2012, 41, 6.
12. Huang, C.J.; Liu, Y.F.; Yin, J.W.; Sun, Y.Q.; Long, S.M. Inversion of Target Surface Masking Function Parameters and
Optimization of BRDF Statistical Modeling. Infrared Laser Eng. 2010, 39, 883–886+891.
13. Gong, L.; Pan, Y.Q.; Wu, Z.S. Study on orientation diagnosis of Rayleigh defect particles in optical surfaces by light scattering
characteristics. Acta Photonica Sin. 2014, 43, 831003.
14. Le Hors, L.; Hartemann, P.; Dolfi, D.; Breugnot, S. Phenomenological model of paints for multispectral polarimetric imaging.
In Proceedings of the Spie the International Society for Optical Engineering, Newton, MA, USA, 31 October–1 November 2001;
pp. 4370.
15. J.L.Saunderson. Calculation of the Color of Pigmented Plastics. J. Opt. Soc. Am. 1942, 32, 727–729.
16. Li, J.W. Infrared Polarization Imaging Technology and Application; Science Press: Beijing, China, 2017.
17. Murphy, B., A. Modified Kubelka-Munk model for calculation of the reflectance of coatings with optically-rough surfaces. J.
Phys. D Appl. Phys. 2006, 39, 3571.
18. Yang, Y.F.; Wu, Z.S.; Cao, Y.H. Practical six-parameter bidirectional reflectance distribution functionmodel for rough surfaces.
Opt. Eng. 2012, 32, 306–311.
19. Liu, Y. Design of Infrared Imaging Optical System for Space-Based Space Targets Based on Micro Polarization. Master’s Thesis,
, Changchun University of Science and Technology, Changchun, China, 2020.