The Key Technology Research On Identification Of Oil Spill On SAR Image

As an important part of the global life system, the ocean has great mediation functionto the earth ecological environment. However, with the rapid development of industry,marine pollution is more and more serious. Especially in recent years, oil spill accidentsfrequently occured, which makes the marine oil pollution problems become increasinglyprominent, and affects the seafood value, other maritime activities and human health.Therefore, monitor the oil spill pollution and protect the marine environment is urgentlyneeded. In recent years, with the development of satellite remote sensing technology,remote sensing has become the most important and the most efficient means on oil spillmonitoring, especially after the successful launch of the synthetic aperture radar (SAR),the capacity on monitoring the sea oil is gradually improved. Actually how to identify themarine oil film based on SAR accuratly is one of the great hot for many scholars.Based on the oil spill SAR image as the object of study, this paper discussed the keytechnique to identify the oil spill, and analysed the feasibility and superiority of the fuzzysegmentation method and the object oriented method in oil film with texture feature. Firstof all, preprocess the test SAR image, including image radiation correction and geometriccorrection and filtering. When filtering, the filter and calculation window were comparedand analyzed, finally choosing the EnhanceLee filtering in3*3window which hasbetter denoising effect and better edge preserving effect. Then, identify the oil film basedon fuzzy theory, using fuzzy threshold image segmentation method, the maximum fuzzyentropy and the minimum fuzzy entropy segmentation algorithm were applied on the oilfilm recognition, finally after the comparative analysis, the2-D minimum fuzzy entropyimage segmentation method is more suitable for oil spill recognition on SAR image.Finally, on the recognition between the oil film and similar oil film, in addition to the grayfeature, this paper used the object-oriented method combined with the four key texturefeature extracted and compared with the traditional methods such as maximum likelihoodmethod, neural network and support vector machines. The results showed that theapplication object-oriented classification method with texture feature for the similar oilfilm have better recognition effect and can be more accurately to distinguish the oil film and similar oil film.In short, the ultimate aim of this paper is to provide a convenient means for themarine oil pollution monitoring, and discover the oil pollution phenomenon such as oillines spill and oil leakage in good time. This will provide technical services for for Marinepollution control, oil and Marine fishery.