3D auto-reconstruction for street elevation based on line and plane feature

3D reconstruction based on high-level features, such as line and plane, is a significant development trend in Digital Photogrammetry and Computer Vision. Moreover how to rapidly realize 3D visualization of street landscape is very important for cyber city's 3D modeling. In order to improve efficiency of street 3D visualization and reduce human-computer interaction, a scheme with high feasible and high automatic approach to auto-reconstruct 3D street elevation is presented by means of line matching, solving orientation elements by vanishing point, auto-rectifying and auto-mosaiking large obliquity-angle close-range sequential images, auto-recognizing concavo-convex edge of street elevation. Our approach for 3D-reconstruction bases on the feature of line and plane, which is also independent of model parameter, vectogram or assistant of space point cloud. Experimental results prove that reconstructed 3D modeling is correctness, effectiveness and feasibleness by our method, which has high degree of automation and should be able to meet the need of rapidly visualizing 3D street elevation landscape.