International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
Homogeneity: 
pr, 
YN dor (i J) (4) 
li- 
4. INTERPRETATION 
4.1 Radar Image Smoothing 
A detailed analysis of the radar image shows that even for a 
single surface type, important grey level variations may occur 
between adjacent resolution cells. These variations create a 
grainy texture, characteristic of radar images. This effect, 
caused by the coherent radiation used by radar systems, is 
called speckle. It happens because each resolution cell 
associated with an extended target contains several scattering 
centers whose elementary returns, by positive or negative 
interference, originate light or dark image brightness. This 
creates a "salt and pepper" appearance. 
The homogeneous patches representing the fields have high 
variability in backscattering due to the speckle noise. This 
results in a grainy image, which renders difficult the 
interpretation of the main features of the surface imaged by the 
SAR. A filer for smoothing noisy radar images is performed. 
4.2 Extraction of Texture Features 
The indistinctiveness and uncertainty of remote sensing data 
due to multiple factors including random factors, the texture 
reflected on remote sensing images are not regular and 
generally do not repeat as cloth patterns. Therefore, texture 
information only has statistical meaning. Statistical. texture 
analysis method is prevalent now. 
Radar imaging has its own specific characteristics that are quite 
different from optical and infrared remote sensing. Radar image 
depends heavily on the scatter of ground objects and its textures 
sharply vary with different objects. And 14 texture features 
could be computed from the co-occurrence matrices. By 
comprehensive analysis and comparison of the 14 texture 
feature, it is founded that uniformity of energy which is a 
measure of image homogeneity, variable which reflects a image 
heterogeneity, entropy which describes the image complexity, 
are more suitable to identifying the inundated area. 
The basic method for water, road and residential arca detection 
is thresholding. A number of threshold levels can be defined to 
separate various ranges of texture value. We choose the value 
located at tough point as the threshold from the histogram of 
texture images. Figure 2 b, c, d respectively represent the water 
area from the below texture images. It was easily found that the 
areas shadowed by mountain were mistakenly detected as 
flooded area. By using the DEM these areas can be 
automatically detected from the derived images. 
Compared with the ground truth, an image interpreted visually 
from SAR data (shown as the contour line of water bodies). We 
can find that the main errors distribute in ramification. Of which 
the result of the extracted water segments using homogeneity 
feature was best ( Yang et al, 1998). 
  
D 
a. Original imag 
= 
   
c. Using ST Deviation feature d. Using homogeneity feature 
Figure 2. Extraction of water texture feature 
Xe 
maio 
T pon 
  
c. Using contrast feature d. Using Skewness feature 
Figure 3. Extraction of road texture feature 
Figure 3 b, c, d respectively represent the road area from the 
above texture images. In general, the combination of using tonal 
features along with textural features improves the correct 
extracting rate over using either type of feature alone. But the 
cement road and river in the SAR images show dark tone. It 
was easily found that the cement road was mistakenly detected 
as river. By using the DEM these areas can also be 
automatically detected from the derived images. 
  
b. Using ST Deviation feature 
a. Original image 
Figure 4. Extraction of residential area texture feature 
Internat 
  
Figure 
residenti 
same te» 
area feat 
43 Im: 
Differen 
characte 
strong t 
water pi 
puts up i 
[n the at 
the test 
residenti 
  
  
In figure 
extractio 
is feasibl