Research On Oil Spill Monitoring By Spaceborne SAR With Texture

Oil spill is one of the pollutions which can bring serious impacts to the marine eco-environment, and attracts the attention of countries and the public. When oil spill accident occurs, it is necessary to determine the location of spilled oil in order to make a quick response and minimize the negative impact of the accident. Since oil spill is unpredictable, monitoring technique should operate in all weather conditions. Compared with aerospace, aviation, and ground, the three remote sensing platforms, spaceborne synthetic aperture radar (SAR) is capable to monitor oil spill timely and correctly, and is selected as the main tool.Oil film suppresses the surface capillary wave, and reduces the roughness of sea surface, which causes a dark area in SAR image. However, other phenomena also take on similar feature as oil film. So, it is necessary to distinguish oil film from other look-alikes. In this paper, texture is introduced in order to identify oil slick correctly.This paper takes "Prestige" accident as the example, and extracts oil spill information by SAR data. Furthermore, the paper also discusses SAR image preprocessing, especially selection of filter methods and filter sizes. With regard to texture features, Gray-level Co-occurrence Matrix (GLCM) is selected, because it considers gray value of pixel and the relationship between pixels at the same time. The paper provides detailed research on texture analysis between oil film and look-alikes, especially on the texture parameter selection, such as direction, step length, and calculation window size. In order to improve the efficiency, the paper further selects the texture feature vectors which are more suitable to oil spill monitoring. At last, three supervised classification methods, the minimum distance, maximum likelihood, and support vector machine (SVM) are compared. It shows that SVM has a better effect for oil film identification with texture features as input. Finally, the paper summarizes oil spill monitoring methods, and brings forward some problems that need to be resolved further.