基于道路信息的高分辨率SAR图像道路检测  

Road Detection in High-resolution SAR Image Based on Road Information

作　　者：刘明飞 张叶枫 付凯城 曲毅[1] LIU Mingfei;ZHANG Yefeng;FU Kaicheng;QU Yi(Department of Information Engineering,Engineering University of PAP,Xi'an 710086,China)

机构地区：[1]武警工程大学信息工程系,西安710086

出　　处：《遥感信息》2017年第6期115-121,共7页Remote Sensing Information

基　　金：国家自然科学基金青年基金(61101238)

摘　　要：针对光学图像道路提取方法不适用于高分辨SAR图像道路提取的问题,提出一种以道路宽度和平行双边缘先验信息为依据的提取算法。算法首先采用各向异性扩散滤波滤除相干斑噪声;其次采用Canny算子和ROA算子进行边缘检测,并对图像边界的信息进行了补偿;而后利用平行线检测方法提取边缘中的平行线特征;最后采用启发式连接的思想连接断裂的线基元。采用算法仿真并对SAR图像进行处理,实验结果表明,该算法对直线和曲线型的道路都具有良好的检测效果。此外,该算法还可以用于提取机场跑道、河流等其他具有平行双边缘的地物。Considering the method of road detection used in optical image is not applicable in high resolution synthetic aperture radar(SAR)image,a road detection method is proposed for SAR image based on the double parallel edge priori information of road.Firstly,the anisotropic diffusion filter is adopted to filter out the speckle noise.Secondly,the Canny operator and ratio of average(ROA)operator is used respectively to get the edge points,and the edge information of the image is compensated.Then,extract the edge with parallel characteristic using the parallel detection method.Finally,the broken lines are connected by means of the heuristic link idea.The algorithm is applied in simulation and SAR image.The experimental results show that the proposed algorithm can extract the road both linear and curve type efficiently.In addition,the algorithm can also be used to extract the ground object with parallel double edge such as airport runway,rivers and so on.

关 键 词：合成孔径雷达 道路信息 Canny-ROA算子 平行线检测 启发式连接 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]