Transcript Document

Sar polarimetric data analysis for
identification of ships
India Geospatial Forum – 14th International Conference
February 07-09, 2012, Gurgaon
S. Swarajya lakshmi
ADRIN, Dept. of Space, Govt. of India
Objectives
• Exploitation of polarimetric SAR data for
detection of ships
• Understanding the scattering mechanisms of
ships through decomposition
• Feasibility for deriving additional information
for identification and classification of ships
Polarization Combinations
VV
HH
HV
VH
pq – p –transmit
q -receive
Polarimetry : Information Content
As compared to single-polarization SAR, polarimetric SAR
provides additional information on:
• Type of scatterer: Trihedral, dihedral, dipole etc.
• Orientation of the scatterer about the radar line of sight
• Ellipticity: degree of scatterer symmetry
• Entropy: significance of the polarimetric information
Therefore, enables better characterization of the target
Scattering Mechanisms
T11= |hh+vv|2
T22 = |hh-vv|2
T33=2|hv|2
Polarimetric Signature
Horizontal Polarization  = 0º or 180º
Vertical Polarization
 = 90º
Pedestal Height
Circular Polarization
Elliptical Polarization
Linear Polarization
Vertical Polarization
Linear Polarization
Elliptical Polarization
Circular Polarization
 = 0º
-45º <  < 0º and
0º <  < +45º
 = -45º or +45º
Materials & Methods
Data Used:
Radarsat 2
Acquisition Type
Fine Quad Polarisation
Product Type
SLC
Date
22-02-2009
Pixel Spacing
4.733m
Swath
25km
Approximate Resolution
Range: 12m
Azimuth: 8m
Incidence Angle
20 – 41 degrees
Software Used: POLSAR of ESA
Methodology
Steps Involved
1. Input SLC data
2. Sinclair Matrix – Shh, Shv, Svh, Svv
3. Extracting Different Target Descriptors –
Stokes matrix, Covariance Matrix, Cherence Matrix
4. Speckle Filtering
5. Polarimetric Parameter Extraction –
Total Power, Entropy, Alpha, Anisotropy, Degree of Polarisation,
Eigen Analysis parameters etc.
6. Extracting Polarimetric signatures
7. Polarimetric Synthesis
8. Polarimetric Decomposition and Classification
9. Separation of Land and Water
10. Identification of anomalies in water
11. Identification of ships
12. Further characterisation of ships with respect to polarimetric
parameters
Entropy
 Eigen Values: Three eigen values of the 3x3
Coherency matrix λi represent the intensities
of the three main scattering mechanisms
 Probabilities Pi of each scattering mechanism
Entropy (H)
This is a measure of the dominance of a given
scattering mechanism within a resolution cell.
Entropy ranging from 0 to 1, represents the
randomness of a scattering medium
from isotropic scattering (H=0)
to totally random scattering (H=1)
3
H    Pi log3 Pi 
i 1
Where,
Pi 
i
3

j 1
j
ENTROPY
Alpha
If the Entropy is close to 0, the alpha angle provides
the nature or type of the dominant scattering
mechanism for that resolution cell.
For example it will identify if the scattering is
volume, surface or double bounce.
anisotropic
odd bounce
 = 0
Isotropic
odd
bounce
 

anisotropic
even bounce
 = 45
Multiple
 = 90
Isotropic
even
bounce
3
1
2
cos1 v11  
cos1 v12  
cos1 v31 



3
P 
i 1
i
i ,
w here
 i  cos1 vi1 ,
0    900
ALPHA
vi1 = first element of the ith eigenvector
Anisotropy (A)
This is the measure of how homogeneous a
target is relative to the radar look direction.
For example, the
Amazon forest is a very homogeneous target
and would have a low anisotropy value.
In contrast, row crops would have a high
anisotropy value.
A
2  3
,
2  3
0  A 1
A indicates the distribution of the two less
significant eigenvalues
Anisotropy becomes 0 if both scattering
mechanisms are of an equal proportion;
values of A > 0 indicates increasing amount of
anisotropic scattering.
Target Decomposition
• Analysis methods whereby individual
scattering
components
that
have
meaningful physical interpretation can
be identified in the received signal.
• Scattering matrix is decomposed into
sub-matrices
so that Individual
component have physical meaning =>
Surface scatterer, double bounce, volume
scattering
H-Alpha Scattering Plane
Classified image depicting water and ships
Anomalies in water
identified as ships
Classified Image
Land
Ships with typical signatures
Class description
Scattering Mechanisms with respect to
the ships identified
Turbulence of water
Boundary between water &
metallic ship body
Ship structure
Objects causing strong double
bounce scattering
Polarimetric signatures
Polarimetric signatures – ship
Proportion (%) of pixels for each class of scattering
Class 1
Ship
2
3
4
5
6
7
8
1
3
9
5
60
21
-
4
1
2
3
4
4
2
4
5
8
12
8
5
6
50
56
56
21
18
15
-
6
6
4
7
5
4
Proportion (%) of pixels for each class of scattering
Derived Information on Ship Measurements
Conclusions &Way Forward
 Typical scattering mechanisms were
observed to be associated with the ships,
which could be used towards automated
detection and characterization of ships.
 Potential of the polarimetric data cold be
further explored with multi-parametric
decomposition schemes and tested with a
wide variety of ships.
THANKS
FOR YOUR KIND ATTENTION