ISPRS, Vol.34, Part 2W2, "Dynamic and Multi-Dimensional GIS’’, Bangkok, May 23-25, 2001 
EXTRACTION OF THE SEA OIL INFORMATION FROM TM AND AVHRR IMAGE 
BY THE METHOD OF FEATURE DATA LINE -WINDOW 
Fengrong HUANG 
Institute of Nautical Technology, Dalian Maritime Univ., Dalian 116026,China Tel:(86)-411-2664232 
Fax: (86)-411-2664232 
E-mail: mellisa001@sina.com) 
Key words: oil spill, remote sensing information extraction, hot area sampling, FDL-W 
Abstract 
The bands of remote sensing of LANDSAT and NOAA is not ideal for catching Sea Spilled-Oil data, which results in the harder on study 
of monitoring the Ocean environment pollution though the spectrum data of Spilled Oil in sea proves the possibility of extracting the 
Spilled-Oil information from TM and AVHRR image. This paper introduced the method of Feature-Data-Line-Window (FDL-W) to auto 
extract the Sea Oil information from TM and AVHRR images. 
Generally, the accidents of the perils of the sea may cause the 
pollution of spilled oil, which is paroxysmal. Therefore the 
prerequisite to response the accident of Spilled Oil is whether 
we can accurately extract, in a short period of time, the 
information of affected region, i.e. range, area, the quantity of 
the spilled oil etc. So the auto-extraction of the Spilled-oil 
information from Satellite Image is the key issue we are very 
concerned about. Because of the development of the Science 
and Technology of Remote Sensing, quickly update the 
Remote Sensing data is no longer the problem. How to solve 
the issue of quantity of Remote Sensing is more important than 
ever before. In this paper, we would like to discuss the thought 
of the quantity of Remote Sensing base on the model of 
information extraction of oil spills from Remote Sensing Data. 
1. Introduction 
The method of Featured-Data-Line- Window (FDL-W) is 
based on the oil spectrum data. According to the variety of oil 
spills from measuring on Sea Surface Experiment, we got the 
spectral characteristic graph of five kinds of oil film and 
seawater. Then by processing the history data of oil spills 
using the multi-band linear enhancement and composition, we 
got the thematic imagine of oil spills information and selected 
the hot spot area. Afterwards we synchro-sampled the hot spot 
area for all bands of TM or AVHRR data, established the oil- 
spilled gray-level database of all bands, drawn the oil-spilled 
FDL. For the relation of spectrum and thickness of the film, we 
adopted the each FDL-Window as a operator to compare with 
the each pixel-gray on the whole imagine one by one and band 
over band. A computer program controls this procedure. If the 
pixel-gray of each band is all contained within their Window the 
pixel will pick up and record, otherwise it will be canceled. So 
at first we must get all the spectrum data and spot image 
feature of all kinds of spilled-oil, and find the spectral bands of 
spilled-oil film by special image processing system. Then build 
a hot-spot area and take a synchronous sample from all 
channels in this area to set up a spilled-oil image Grey 
intensity database. Finally, make s model of the correlation 
between spectrum value of spilled-oil and the thickness of oil 
film by analyzing spectrum data of spilled-oil. And use 
distinguishing model technology to find out discriminating 
function, then calculate spot image of all channels to extract 
the information. 
2. The Spectrum Feature and Image gray of Spilled-oil 
The sea spilled-oil spectrum data measuring on Sea Surface 
show that after seawater is polluted by spilled-oil its reflectivity 
will have a great change. Usually there is a drop tendency in 
the visible light section whose reflectivity is 0.005-0.033 and 
whose peak value is near 0.55-0.57. The reflectivity in red light 
and reflected infrared section (0.7 -1.040mm) is also low. It’s 
often less than 0.01, but most of it is less than 0.005. The