The Research On Fast Segmentation Algorithm In SAR Oil Spill Images Based On Watershed

Synthetic aperture radar (SAR) is a radar image by microwave reflects signals. It has the characteristics that can work all-time and around the clock, so used widely in the military field. In recent years, with the development of science and technology, SAR becomes more popular and applied more widely. Recently frequent offshore oil spills happened; SAR has played a significant role in the detection of the oil spill region.Due to the factors of SAR images’imaging characteristics and the complex sea conditions, there’re very serious noise problems and uneven grayscale, which brings a lot of interference in SAR images’segmentation. Firstly, this paper studies on SAR images’enhancement method and then focus on a fast segmentation algorithm for SAR images. And then, this paper presents a fast SAR oil spill image segmentation algorithm based on block threshold and watershed, which has characteristics of segmented quickly and high accuracy. And this algorithm solves the problem of that the existing segmentation algorithm does not has actual engineering applications. This paper’s main research work includes following aspects:(1) The research on SAR images’enhancement method. The target in SAR image is generally divided into two kinds, one is point target, and the other is massive target. Target in the SAR oil spill image is dominated by massive target. The result can be seen, SRAD (Selective Replication-based Adaptive Data Delivery Scheme) can effectively remove the speckle noise in SAR oil spill image, has a good smoothing effect in massive target, meanwhile can keep the edges and the target feature well.(2) The research on SAR images’segmentation algorithm. For SAR images’ segmentation method, this paper firstly introduces the conception of block threshold. Dividing an image into four parts, each one uses threshold transform automatically, and then merger these four small images. This can effectively solve the problem of SAR oil spill’s uneven grayscale. Secondly, extracting the foreground and background with corrosion and reconstruction in morphological operations. Finally, getting the image segmentation results of continuous boundary with the watershed.Experiment a large number of SAR oil image in laboratory, this algorithm can segment the oil spill and sea area quickly and efficiently, and has more practical significance in engineering applications.