Since roads play an important role in many application areas, up-to-date road databases are critical. Automated road extraction using many different types of remote sensing data has been explored. Recently, lidar data has proven advantageous for road extraction and researchers have extracted roads from lidar data alone as well as fused with passive imagery. This paper considers techniques to extract roads from lidar data exclusively and explores the impact of land use on the accuracy of the derived road network. To support this goal, roads were extracted from lidar data for six study sites--three residential and three commercial--within Oneida County, NY. The accuracy of generated results was computed and comparison tests were performed. These results showed high levels of accuracy for the raster road cluster delineation, with lower accuracy for the vector centerlines. The comparative analysis showed that land use was a factor in road delineation accuracy for both the raster and vector stages of analysis.
展开▼
