基于数学形态学和相位编组SAR影像道路自动提取  

Automatic road extraction from SAR images based on mathematical morphology and phase group

作　　者：窦建方[1] 陈鹰[1] 

机构地区：[1]同济大学测量与国土信息工程系/遥感与空间信息技术研究中心,上海200092

出　　处：《测绘科学》2009年第2期53-54,63,共3页Science of Surveying and Mapping

基　　金：国家科技支撑计划(2006BAJ091300)

摘　　要：由于SAR图像中固有的斑点噪声的强烈影响,某些对于光学图像有很好效果的边缘检测算子,但是对于SAR图像来说,效果很差,本文应用一种改进的基于图像统计信息的局部线特征检测方法,首先利用图像统计信息对图像进行局部特征检测,然后利用数学形态学去除内部空隙并细化得到道路段,将道路段进行相位编组,进而自动提取出道路。According to the impact of inherent speckle of SAR images, some edge detection operators have a good effect on optical images, but have a poor effect on SAR images. In this paper, an improved local linear feature detecting method, which is based on the statistical information of images, is applied. Firstly, it detects the local linear feature of images with the statistical information. Then, based on mathematical morphology, it removes internal gaps and obtains road sections after thining. With road section grouping, road extraction is performed automatically.

关 键 词：SAR图像 边缘检测算子 线特征 数学形态学 相位编组 细化 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]