Integration of images and laser scanning data for automated 3D road feature extraction

The authors present an automated 3D road feature extraction technique that integrates aerial images with both airborne and mobile laser scanning data. The methodology uses complementary information from images and laser scanning data to improve the automation capabilities of them individually. A bare-earth model of the road corridor is extracted using the random sample consensus method by exploiting the planar nature of road features, such as the sidewalk, median, and road pavement. Then, by using images and laser scanning data, the road features sidewalk, median, and road edges are extracted in 3D. The methodology is validated experimentally by using aerial images with both land-based mobile and airborne laser scanning data, proving that it is a feasible way of generating 3D road models for surveying and transportation applications.