Istanbul 2004 
e can see the 
the software 
| Stech (2003) 
diverse tested 
ccurring small 
1sed statistical 
  
Figure 2, went 
> oceanic area 
  
ge 
diverse white 
nflict in the 
; false targets. 
tablished by 
to have been 
erated by the 
Rayleigh and 
;, without the 
ccurred with 
ll distribution 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
  
  
Figure 4. Weibull distribution processed image 
Already in the Figures 5 and 6 we can see the results obtained 
by the processing based, respectively, in the use of the 
distributions Gamma and Root of Gamma. We can observe that 
these distributions were the ones that presented a smaller 
answer so much for the oceanic area as for land area, mainly the 
distribution Root of Gamma, without, however, to leave of 
detecting the existing targets. 
The Figure 7 shows the result obtained by the processing based 
on the use of the methodology described by Eldhuset (1996). 
This methodology presented also resulted very good, getting to 
detect the existing targets, without a high answer of the land 
area, presenting, however, some noises next to the detected 
targets. 
Figure 5. Gamma distribution processed image 
  
Figure 7. Eldhuset's methodology processed image 
4. CONCLUSION 
We can conclude with this work that the use of statistical 
distributions for segmentation of SAR images that enclose 
oceanic areas with the intention of ships detect is very efficient. 
It was looked in this work to use the more cited statistical 
distributions in existing literature, being only accomplished the 
necessary adaptations for the obtaining of the best possible 
resulted. We can conclude that the use of the distributions 
Weibull, Rayleigh and Gama were the ones that presented the 
best results in relation the others used distributions. As already 
says previously, distribution K and the methodology described 
by Eldhuset (1996) presented a result with some noises next to 
the targets detected, what could make difficult the image 
interpretation because it can be interpreted as false targets. The 
distribution Root of Gamma presented a result whose value of 
threshold established by software for this distribution made 
with that the targets were detected with a smaller size in 
relation the others distributions, fact this that can cause a loss of 
information and, consequently, the omission of some target or a 
wrong evaluation of the targets. We can also conclude that the 
targets were detected with different sizes in accordance with 
each methodology, Such fact, however, does not harm the 
interpretation of the results and the consequent detection of the 
targets. This occurs by virtue of the value of threshold 
established by the software to be different for each distribution. 
References: 
Eldhuset, K., 1996. An automatic ship and ship wake detection 
system for spaceborn SAR images in coastal regions. /EEE