The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B6b. Beijing 2008 
filter results automatically. As the lightness of pixels in water 
region is very low, first we can set the pixels which lightness 
level 128) to background lightness 255; and then employ the 
region grow algorithm to mark all the regions in the result 
image. 
For automatic selecting the water targets, we need calculate 
some characteristic of the marked regions. In this paper, we 
calculate the area value, average gray level and the histogram of 
image and knowledge of water object, we build up the water 
objects model with the following rules: 
(1) Area rule: regard the region which area less than some 
specific value as non-water region. 
(2) Lightness rule: regard the region which average lightness or 
as non-water region. 
(3) Histogram distribution rule: as the water region consist of 
most low lightness pixels and some noise pixels, its histogram 
should has a single peak in the left side. Therefore, we build the 
histogram distribution rule regarding the region can not accord 
with all the following conditions as non-water region: (a) gray 
level of the histogram peak less than the average gray of region; 
(b)the number of pixels with gray level of the histogram peak 
more than 10 percent of total pixels in the region; (c) the 
number of pixels with gray level ranging from the histogram 
peak to the peak level plus 5 more than 60 percent of total 
histogram peak level less than 1 percent of total pixels in the 
Based on these rules and the characteristic of regions, we can 
select the water objects automatically. The Figure 1 is the 
flowchart of the approach proposed in this paper. 
over some specific level (in this paper we select the middle 
every marked region. And based on the characteristic of SAR 
the lightness of the peak of histogram over some specific value 
pixels in the region (d) the number of pixels darker than the 
region. 
Figure 1. Flowchart of the proposed approach 
4. EXPERIMENTAL RESULTS 
We do experiments using the Wujiang SAR image which has 
water region, airport region and mountains. The result proved 
that the method proposed in this paper is feasible and valid, and 
the water object in the SAR image can be extracted rapidly and 
efficiently. Figure 2 is the original SAR imagery region (512 x 
512). Figure 3 is the result image after the sequential nonlinear 
filtering. Figure 4 is the result after water object selection 
automatically. Figure 5 is the water objects extract by algorithm 
proposed in this paper.