33e «009 
23 203 see wor 
„PEAK Loca pL ico 
motor. M 
ario > 5 
Ph "rr 704500 +009 
= 
e "a 3: 0190 
Af 308 Mid 3.0308 n 
-MEASUDED -INTEQ, PU, - 
PAIN Lob, = Hes 
31DE Lobsz 114.92 
1:08 = -5.501 
-( in O07 992-10, - 
MAIN Lob.= 119.04 
SIDE Lobaz 115.94 
= -3,0992 
by the point 
o electromag- 
ommonly hor- 
;,hogonal com- 
four measure- 
the scattering 
ene. The re- 
s have strong 
ncident wave. 
7 polarimetric 
on and classi- 
rimetric mea- 
tering matrix 
r an antenna 
ve, the polar- 
ermined from 
rbitrarily po- 
xpected SAR 
Performing 
sults in image 
tive function 
SAR systems 
enna orienta- 
entations can 
ited, allowing 
images. 
ng polarimet- 
  
  
  
  
  
POLARIZATION SIGNATURE 
INTENSITY 
… 
a 
    
ORIENTATION 
INTENSITY 
  
Figure 13: Polarization signatures of an ideal reflector. 
ric behaviour of specific scatterers is the polarization 
signature [7]. A polarization signature consists of a 
three-dimensional plot of backscatter measurement (ul- 
timately the radar cross section) for a particular dis- 
tributed area as a function of the elipticity and orien- 
tation angles of the transmit antenna. The receive an- 
tenna can be polarized either the same (copolarized) or 
orthogonal to (cross polarized) the transmit antenna. 
Polarization signatures of an ideal reflector as gener- 
ated by EV-SAR are shown in Figure 13. These signa- 
tures are convenient for exhibiting scattering behaviour 
of differing surface areas. 
5.3 Poincaré Sphere 
Another useful tool for observing differences in scatter- 
ing behaviour is the Poincaré sphere. For an arbitrary 
transmit antenna orientation the polarization of the re- 
flected wave can be characterized by a set of quantities 
known as Stokes parameters. The Stokes parameters 
are similar in definition to spherical polar coordinates 
and therefore lend themselves to a spherical represen- 
tation. The Stokes parameters of many scatterers can 
be mapped onto the surface of a sphere, such that a 
given location corresponds to a particular polarization. 
EV-SAR provides the ability to interactively map the 
polarimetric information of chosen scatterers on the 
surface of the Poincaré sphere (Fig. 14). The transmit 
antenna polarization can be specified, and the sphere 
can be rotated to provide a view from any direction. 
6. MULTISENSOR DATA FUSION 
Since SAR measures the backscattering coefficient at 
  
Figure 14: Display of scatterers on the Poincaré sphere. 
RF frequencies (greater than 1.0 GHz), the imagery is 
complementary to many other types of remote sensing 
data. Multisensor data fusion describes the range of 
techniques that are used to combine imagery from 
different sensors into one image for interpretation. 
6.1 Image Coregistration 
For fusion of any two data sets, there must be a map- 
ping between the pixels in one image to those in the 
other. Image coregistration exploits this mapping to 
exactly line up the two images so that they overlay per- 
fectly. Two methods of coregistration are possible with 
EV-SAR. First the images may be coregistered manu- 
ally using ground control points (Section 2.3). This is 
very accurate for small scenes where the respective im- 
age geometry mapping is approximately linear. Second 
method is to simply use the latitude/longitude tagging 
of each image to resample one image (the slave) to be 
coregistered to the other image (the master). This sec- 
ond method is the only possible solution when the slave 
image data has a far lower resolution as compared to 
the master. 
6.2 Grid, Vector, and Point Overlay 
Multisensor data fusion in EV-SAR allows overlay of 
various types of non-raster data. Latitude/longitude 
grids may be overlayed as a geographical reference. 
Vector data such as coastlines extracted from the 
Digital Chart of the World may be overlayed for 
reference and to verify the georeferencing. Other 
types of vector data such as point data samples of 
305