Research On Classification Of The Marine Oil Film Based On SAR Image

Oil is an important fossil fuel, which is called industry blood. With the exploitation of the oil resource, ocean water pollution problems have become more serious. In a variety of different kinds marine pollution, oil pollution tops the list, which has harmed on people’s production and life. Due to the oil spill always occurred in bad weather conditions, Synthetic Aperture Radar(SAR) as an all-weather, all-time, high-resolution active microwave imaging sensor, has been the primary mean for the oil spill detection.This paper studies on oil spill SAR image, and explores the whole process of ocean film identification, by comparing the effects of different methods. Firstly, the SAR image is pretreated, including radiation correction, geometric correction and filtering. On this basis, the sobel operator, seed filling and a single threshold methods are used for image segmentation, and a simple comparison of the advantages and disadvantages of the three segmentation methods, and a simple comparison of the advantages and disadvantages of the three segmentation methods is given. The key part of this paper is on the oil spill image classification. Firstly, the general supervision classification methods are used for image classification. Secondly, GLCM is used for texture information extracted, and then using statistical analysis to filter out useful texture information to support the image classification. Thirdly, the classification accuracies before and after using texture information are compared. The result proves that supported by texture image classification accuracy is superior to the general supervision classification. This paper introduces object-oriented classification to SAR oil spill image classification. The object-oriented image classification is conducted by ways of ENVI EX and eCognition, and effectively identified the oil spill area. The two software systems are compared in doing object-oriented classification.In short, the ultimate aim of this study lies in providing an effective and convenient method for the identification of marine oil spills, by way of comparing different methods and softwares, in order to service for the marine environmental protection and sustainable development strategies.