Study On Marine Oil Spill Information Extraction Using ASAR Imagery

There will be subsequent oil efflorescence and diffusion after an oil spill incident because of the complex maritime environment. The influence of the oil spill incident to the maritime environment and resources will be destructively if monitor and emergency measures are not taken timely. Synthetic Aperture Radar can detect the oil spill according to the different backscatter intensity on the sea with the microwave band. SAR can meet the accuracy demand of detecting the oil spill object with the characteristics of weather proof, high detection accuracy and large coverage area.There are two main problems in researches related to SAR, one is the effect of the look-alike to the accuracy of oil spill detection, the other is lack of space information analysis which hinders target detection from the perspective of space similarity.In this paper, a new information extraction method is proposed, which combined expert knowledge and object-oriented classification method to reduce the effect of look-alike. Simultaneously, textural features are input as classification object, Object-oriented classification method are used to dig the two-dimensional space character deeply. The main innovative points of this paper can be concluded as following:on one hand, classification rules for look-alike of marine oil-spill according to the cause, characteristics and trend of the look-alike of marine oil-spill image are established and the characteristics of oil-spill object of marine oil-spill image and background information of scene of incident are combined to form expert knowledge database to classify and reject the look-alike. On the other hand, considering the limitation of current research, this paper designed a scheme of marine oil-spill image extraction and monitoring combining the object-oriented classification method and look-alike recognition method with consideration of the two-dimensional space character, which can achieve better result of oil-spill target recognition.This paper used the aforementioned method and technology, taking the example of the marine oil-spill incident caused by Lebanon war in2006, with data from ENVISAT-ASAR got the conclusion that compared to the non-knock out method, look- alike expert knowledge database combines with the object-oriented classification method can knock out suspected oil-spill object which improve the efficiency and accuracy of the classification algorithms greatly.As an important tool for environment disaster monitoring, SAR get attention from the world and shows characteristics like constellation-allied, wave brand diversified and multi-polarization. The above characteristics permit SAR to provide better data for oil-spill information extraction scheme, which will improve the accuracy of the detection of the oil-spill object.