任意方向对称差值核的SAR图像边缘提取算法

发布时间：2018-01-21 02:32

本文关键词： SAR图像边缘提取对称窗口核函数　出处：《测绘学报》2017年09期 　论文类型：期刊论文

【摘要】：由于现有边缘检测方法多存在对图像噪声的敏感性以及检测边缘的有限方向性等问题,因此在斑点噪声的影响下难以有效检测SAR图像中具有任意方向性且强度相近的区域间边缘。本文提出了一种任意方向对称窗口的SAR图像边缘提取算法。首先,以任意像素在图像中的格点位置为对称中心构建任意方向上的对称窗口;定义各窗口内像素到中心像素距离的核函数;以像素核函数为权重分别计算两对称窗口内像素光谱测度的加权平均值,并定义两者之差的绝对值为该方向上该像素隶属边缘的强度;选取所有方向中最大边缘强度作为边缘像素的判别标准。为去除错误提取的边缘像素,设计基于滤波操作的后处理过程。对模拟图像提取结果的定性及定量分析表明提出算法可在一定程度上克服噪声影响,较为准确地提取任意方向的边缘。
[Abstract]:Because the existing edge detection methods have many problems such as sensitivity to image noise and limited direction of edge detection and so on. Under the influence of speckle noise, it is difficult to detect the edges between regions with arbitrary direction and similar intensity in SAR images. In this paper, an algorithm for edge detection of SAR images with arbitrary directional symmetric windows is proposed. First. A symmetric window in any direction is constructed with the lattice position of any pixel in the image as the symmetry center. Define the kernel function of the pixel distance from the pixel to the center pixel in each window; The weighted average value of pixel spectral measure in two symmetric windows is calculated with the pixel kernel function as the weight, and the absolute value of the difference between the two is defined as the intensity of the pixel belonging to the edge in this direction. The maximum edge strength in all directions is selected as the criterion of edge pixels. The qualitative and quantitative analysis of the simulated image extraction results show that the proposed algorithm can overcome the influence of noise to some extent and extract the edges in any direction accurately.
【作者单位】：辽宁工程技术大学测绘与地理科学学院遥感科学与应用研究所;
【基金】：国家自然科学基金(41271435);国家自然科学基金青年基金(41301479) 辽宁省自然科学基金(2015020090) 辽宁省教育厅科学技术研究一般项目(LJCL009)~~
【分类号】：P237;TN957.52
【正文快照】： 合成孔径雷达(synthetic aperture radar,SAR)具有监测范围广、分辨率高、全天时、全天候等优点,已被广泛应用于遥感、军事和水文等领域[1-3]。目前SAR图像处理算法开发及其应用日益成为遥感学界的研究难点和热点。边缘检测是提取图像中光谱测度跳变特征的过程,也是实现图像分

【相似文献】