Road Extraction from High Resolution Multi Aspect Sar Images

In this paper, we propose a fusion strategy for extracted roads from multi-aspect SAR images. The fusion strategy extends a system for automatic road extraction from SAR images based on line extraction and explicitly modeled knowledge, which has been developed for single SAR images. Due to the side-looking geometry of SAR, the visibility of roads is often limited by adjacent high trees or building rows. Roads in viewing direction (range) are less affected by shadow and layover effects from neighbouring objects than roads across the viewing direction (azimuth). In the fusion step, road segments running in range direction, “good candidates for roads” are higher evaluated than roads in azimuth direction. The fusion technique was tested on a sub-urban SAR scene. The results show the potential of the proposed fusion strategy and the usage of multi-aspect SAR views in case of road extraction.