海洋SAR图像分割及边缘梯度特征的提取

Marine SAR image segmentation and edge gradient feature extraction

为提高合成孔径雷达(SAR)海洋溢油图像的分割效果,得到海洋表面溢油的准确信息,提出一种改进的迭代算法,能更为有效地选取阈值,尤其适用于暗斑区域与背景灰度值差异悬殊的情况。对海洋SAR图像进行预处理,并且对处理后的SAR样本图像进行边缘分割,在此过程中通过优化迭代算法得到一种新的选取阈值的方法,运用此方法成功将暗斑区域与海洋背景分离;使用5*5窗口提取样本图像的边缘梯度特征量,对梯度均值与方差进行分析、理解达到识别SAR图像中的溢油区域的目的。将仿真结果与传统分割结果进行了比较,比较结果表明了该方法在选取阈值进行图像分割方面要优于其它算法。

To improve the sea oil image segmentation effect of synthetic aperture radar（SAR）and get the precise information of ocean surface oil spill,an improved iterative algorithm is presented to select the threshold more accurately and effectively.It＇s especially suitable for situation where the dark spot area＇s gray value differ greatly from background.Firstly,the SAR image segmentation is performed after preprocessing on the basis of improved iterative algorithm to separate dark spots from background efficiently.Then,a 5＊5 window is used to extract the image gradient feature to recognize the oil spilling area in the SAR image by gradient analysis.Finally,the superiority of the optimization algorithm to other algorithms is demonstrated by comparing the simulation results with the traditional segmentation results.