Geometric modeling of urban scenes from satellite imagery
- Publication Date
- Apr 21, 2017
- External links
Automatic city modeling from satellite imagery is one of the biggest challenges in urban reconstruction. The ultimate goal is to produce compact and accurate 3D city models that benefit many application fields such as urban planning, telecommunications and disaster management. Compared with aerial acquisition, satellite imagery provides appealing advantages such as low acquisition cost, worldwide coverage and high collection frequency. However, a set of technical challenges are imposed by the physical constraints of satellite acquisition. Existing methods produce, at best, dense Digital Surface Models of varying degrees of fidelity. This thesis proposes a fully automatic pipeline for producing compact, semantically-aware and geometrically accurate 3D city models from stereo pairs of satellite images.The proposed pipeline relies on two key ingredients. First, geometry and semantics are retrieved simultaneously providing robust handling of occlusion areas and low image quality. Second, it operates at the scale of geometric atomic regions which allows the shape of urban objects to be well preserved, with a gain in scalability and efficiency. Images are decomposed into convex polygons via Voronoi diagram, with geometric regularities captured by aligning contours of atomic polygons with linear structures. Radiometric semantic classes, elevations, and 3D geometric shapes are retrieved in a joint classification and reconstruction process, interacting information from the left and the right images in a satellite stereo pair.Experimental results show the scalability, time-efficiency, robustness, and output quality of the proposed approach. Worldwide cities in different styles are automatically reconstructed in a few minutes. Qualitative and quantitative evaluations demonstrate the fidelity of the output LOD1 representations by comparing with LiDAR solutions and traditional DSM methods. The proposed method generates compact 3D models within a reasonable average error to the ground truth, illustrated by an altimetric accuracy measurement on Denver city.