一种高分辨率SAR影像道路提取方法  

Amethod of road extraction for high resolution SAR image

作　　者：刘振宇 余洁[2,3] 

机构地区：[1]中南民族大学资源与环境学院,武汉430074 [2]首都师范大学城市环境过程与数字模拟国家重点实验室培育基地,北京100048 [3]首都师范大学资源环境与旅游学院,北京100048

出　　处：《测绘科学》2017年第11期107-112,120,共7页Science of Surveying and Mapping

基　　金：国家自然科学基金重点项目(41130744);中央高校基本科研业务费专项资金项目(CZQ14020)

摘　　要：针对高分辨率SAR影像中细节信息损坏道路的面特征结构、影响道路提取,基于影像统计特征,给出一种结合区域生长和细节信息识别的道路提取方法。该方法通过区域生长提取呈现面特征的暗目标(道路框架),利用CFAR算法识别细节,通过形态学融合得到最终结果。为降低高分辨率影像区域异质性对提取结果影响,提出了一种自适应CFAR算法,相比之前算法可自适应删除干扰点;并引入有效表征影像统计的GA0分布。利用海南省陵水黎族自治县机载X波段高分辨率SAR数据的幅度影像进行实验,结果表明,该方法能有效提取呈面特征的道路,获得准确的道路宽度和中心线信息。Aiming at the problem that road surface feature is damaged by details of images which make difficult for road extraction,a method for high-resolution SAR images extraction is presented based on the statistics information of images.The dark surface targets（road frame）are assembled by region growing,the details are detected by CFAR algorithm,and final result is the fusion of the two above by morphological integration.For reducing the heterogeneity of high resolution images,an adaptive CFAR algorithm is proposed.Compared to other algorithms,the interference data can be adaptively removed in the proposed algorithm.And GA0 distribution which can represent the statistical distribution of data is also introduced.The amplitude images of airborne high-resolution X-band SAR data acquired in Hainan Lingshui County are used to validate the method.Results show that this method can effectively extract road of surface feature,accurately obtain the road width and centerline.

关 键 词：高分辨率SAR影像 道路网提取 CFAR GA0分布 

分 类 号：P237[天文地球—摄影测量与遥感]