基于混合模型的SAR影像海陆分割算法

发布时间：2018-05-05 03:00

本文选题：合成孔径雷达(SAR) + 海陆分割　；参考：《光电子·激光》2017年03期

【摘要】：合成孔径雷达(SAR)影像的海陆分割是诸如海洋目标检测和识别等基于海洋区域SAR影像解译的基础和关键环节之一。为解决复杂背景下遥感影像海陆分割问题,提出一种基于混合概率模型的海陆分割算法。首先利用Harris角点检测算法检测出影像中包含角点的图像子块,进而通过均值漂移(MS,mean-shift)算法对图像子块进行聚类分析得到陆地区域的像素样本;然后利用陆地的像素样本,通过最大期望(EM,expectation maximization)迭代算法拟合出混合模型概率密度分布的相关参数;最后通过混合概率模型检测出陆地前景区域,得到海陆分割结果。实验证明,本文方法能够对包含海陆的异质遥感影像实现有效的海陆分割。
[Abstract]:Sea land segmentation of synthetic aperture radar (SAR) image is one of the basic and key links in Ocean region SAR image interpretation, such as ocean target detection and recognition. In order to solve the sea land segmentation problem of remote sensing images in complex background, a sea land segmentation algorithm based on mixed probability model is proposed. First, the Harris corner detection algorithm is used to detect the sea land segmentation algorithm. The image subblock which contains the corner points is detected, and then the pixel samples in the land area are obtained by means of the MS (mean-shift) algorithm, and then the correlation parameters of the probability density distribution of the mixed model are fitted by the maximum expectation (EM, expectation maximization) iterative algorithm using the land pixel samples. In the end, the land foreground region is detected by the mixed probability model, and the results of sea land segmentation are obtained. The experiment proves that this method can achieve effective sea land segmentation for heterogeneous remote sensing images containing sea and land.

【作者单位】：河南大学图像处理与模式识别研究所;河南省空间信息处理工程实验室;
【基金】：国家自然科学基金(61305042) 中国博士后面上基金(2015M582182) 河南省科技攻关(152102210057) 河南大学基金(2012YBZR005)资助项目
【分类号】：TN957.52

【相似文献】