基于多特征的SAR影像溢油暗斑提取

由于星载合成孔径雷达(Synthetic Aperture Radar,SAR)溢油影像包含大量斑点噪声,仅依靠传统模糊聚类方法不能有效提取出其溢油区域。针对SAR图像存在的斑点噪声问题,本文提出了一种结合多特征与改进模糊C均值聚类(Fuzzy C-means,FCM)的溢油暗斑提取方法。该方法首先提取影像的多个特征,以便更加充分地反映影像信息;然后同时考虑像素与其邻域的强度和空间位置关系,以此来构造模糊加权因子,进而定义目标函数;最后通过迭代最小化目标函数,获得最佳溢油暗斑提取结果。文中对真实的SAR溢油影像进行了溢油暗斑提取实验,并分别与利用单一特征和加入邻域关系的模糊聚类方法得到的提取结果进行对比分析,实验结果证明了本文方法的有效性。

Feature Extraction of Oil Spill Dark Spot Based on Multi-feature in SAR Image

According to the characteristics of the Synthetic Aperture Radar(SAR)oil spill image contains a large number of speckle noise, traditional fuzzy clustering method cannot effectively extract the oil spill area, Aiming at the problem of speckle noise in SAR image ,this paper proposes a combination of multi-feature and improved fuzzy c-means ( FCM) method which for extracting oil spill dark spot. The method firstly extracts multi-feature of the image to reflect the image information more fully. At the same time, it con-siders intensity relation and spatial position relation between the pixel and its neighborhood to construct the weighted fuzzy factor, and then to define the objective function. Finally, the optimal result of the optimal extraction of oil spills and dark spots is obtained by iter-atively minimizing the objective function. In this paper, the extraction experiment of real SAR oil spill images are carried out and the results are compared with the single feature and the fuzzy clustering method which is added to the neighborhood relation, the experi-mental results verify the effectiveness of the proposed algorithm.