ISPRS, Vol.34, Part 2W2, "Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
135 
il oil 
;ness of it 
do with 
film. The 
5 a series 
i along Y 
window in 
i alterable 
led-oil and 
data line 
W). There are three kinds of mJDegree, they are normal 
degree, mediate degree, wide degree. The image output of 
each kind of oil is different appreciably. We can learn from it 
that first the method of Featured-Data-Line- Window (FDL-W) 
is only correlated with the thickness of oil film. The different 
kinds of mjdegree are used to search different thickness of oil 
film. Second the light diesel oil can diffuse quickly on the 
surface of the sea and form an even oil film easily. So the 
result of three kinds of mjdegree is same basically. The result 
of other kinds of oils is different from light diesel oils’. Crude oil 
has obvious contrastive result. As crude oil will turn into solid 
when it is in cold in the surface of sea, the thickness of oil film 
is very different. 
To sum up, whether theory analysis or the results of image 
process proves that the method of Featured-Data-Line-- 
Window (FDL-W) can do the work of Auto-extracting the Sea 
Oil Information. The accuracy of it depends on the size of 
window that is related to the kinds of oil and thickness of oil 
film. If the correlation between thickness of oil film and 
spectrum value is great the size of window will have a huge 
influence on the results of extraction otherwise the size of 
window will have a little influence on the results of extraction. 
This correlation can be gotten from spectrum feature 
database. The accuracy of information extraction depends on 
the Window. Actually, the size of the Window is related to what 
kind oil is and how thick oil film is. The more relevance 
between spectrum data and thickness of oil film, the more 
important between choosing the size of the window and the 
result of the extraction. Otherwise, the size of the Window 
won’t affect the result much. Therefore, choosing the size of 
the window is complicated. We can find the relevance from 
the database of the spectrum characteristic. 
So which kind of the size of window should to be selected is 
an important and complex problem. The advantage of this 
method is its reliability and convenience. It collects information 
from all channels, and the information can be recognized only 
all criteria are met. One thing should bring our attention, which 
is the pure of sample of oil spills, and the edge pixels result 
from the spatial accuracy etc. 
Reference 
1. Guo Defang: 1987. The Computer Process and Mode 
Distinguishing of Remote Sensing Image, Electronic and 
Industry Press, 
2. Li Qijun, Huang Fengrong: 2000,Study on Marine Optics 
Remote Sensing and Spilled-oil Satellite Remote Sensing 
and Monitor Technology, 
3. Ma Ainai: 1997. Remote Sensing Information Model, Beijing 
University Press, 
sure if the 
>rmation of 
mation and 
iry function 
and output 
hat are 1 df 
Ne process 
ndow (FDL-