Figure 1: A SAR image before (left) and after (right) 
antenna pattern removal. (SAR image was produced 
by the CCRS C/X Airborne SAR, courtesy of CCRS, 
NRCan, Canada.) 
face height relative to the average Earth radius, near 
range, and far range. 
2.3 SAR Simulation and Geocoding 
Geocoding (a SAR image) is the process of trans- 
forming a SAR image into an image whose coordinates 
are based on a geographical coordinate system, such 
as Universal Transverse Mercator (UTM) or latitude- 
longitude sampling grids. 
SAR images of mountainous terrain are very sen- 
sitive to the terrain elevation. The range-dependent, 
geometric parallax found in these images makes it dif- 
ficult to accurately interpret the shapes of topograph- 
ical features. Significant variations in image gray level 
associated with geometric foreshortening confuse and 
in some instances completely mask image contrasts as- 
sociated with changing surface cover. These phenom- 
ena make it difficult to understand and interpret such 
scenes, especially for interpreters trained in optical im- 
age analysis. 
To correct for the induced terrain based image 
distortions, SAR imagery of irregular terrain elevation 
is geocoded. 
The EV-SAR geocoding algorithm was adapted from 
the geocoding algorithm and software developed by 
Bert Guindon of the Canada Center for Remote Sens- 
ing [2]. It is based on prior knowledge of the terrain (in 
the form of Digital Terrain Elevation Data (DTED) or 
Digital Elevation Model (DEM)) and the parameters 
Figure 2: A section of an ERS-1 SAR image near 
(119E,50N). The high relief resulted in noticeable 
layover. (©ESA 1991) 
of the SAR platform. In the process of geocoding, a 
simulated SAR image is generated. This image is then 
co-registered with the real SAR image of the same ter- 
rain. Using this mapping and the terrain-to-SAR rela- 
tionship already established, a geometric transforma- 
tion is calculated, and the real SAR image is remapped, 
which produces the elevation-effect corrected image. 
Optionally, other useful auxiliary data such as 
shadow and layover masks can also be generated (these 
characterize the interaction between the SAR viewing 
geometry and the local terrain orientation). 
Figures 2 through 5 show some of the data and result 
images in an example geocoding session. 
3. SAR IMAGE ENHANCEMENT 
Coherent processing of SAR data makes images 
sensitive to small range variations on the order of the 
radar wavelength. These variations appear as speckle 
noise (manifested as an apparently random distribution 
of conspicuously bright or dark pixels) in the SAR 
image. The presence of speckle in an image reduces 
the ability to resolve fine details. It also degrades 
classification accuracy. EarthView provides a number 
of tools for the removal of speckle noise, as well as other 
types of noise. Images can be further enhanced using 
EarthView through histogram modification as well as 
classification. 
3.1 Speckle Filters: Frost, Lee, MAP, MAP-Refined. 
300 
  
  
Figure 
the SA 
lat-long 
  
Figure 
the D' 
platfor