期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2005
卷号：XXXVI-8/W27
出版社：Copernicus Publications
摘要：In the last years a system for automatic road extraction from SAR images based on line extraction and explicitly modeled knowledge has been developed at the Technische Universitaet Muenchen (TUM). In this paper, this approach is extended towards the use of multiple views from different viewing directions. The visibility of roads in SAR images is often limited by neighboring tree or building rows on one or on both sides of the road. Only roads in viewing direction are less affected by shadow and layover effects from neighbored objects. Therefore, we introduce a fusion technique where each road segment is assessed according to its direction to the SAR sensor. Road segments running in along-track direction are higher evaluated in contrast to road segments running in across track. The proposed fusion technique is demonstrated on a sub-urban SAR scene. The scene is viewed by three different aspect angles. The result shows the potential of proposed fusion strategy: with the use of three views, the resulting road network is a more complete and more correct than for each single image
