Extracting Wire-frame Models of Street Facades from 3d Point Clouds and the Corresponding Cadastral Map

This paper presents an operational approach for extracting wire-frame models of street facades. Georeferenced terrestrial laser rawdata are acquired by a Mobile Mapping System (MMS). The MMS constitutes an adapted tool to massively collect accurate data at street level. After a filtering step of the 3D raw point cloud of street, the street point cloud is segmented into a cloud of dominant facade walls. Piecewise quasi-planar facade cloud are then extracted using an adapted Progressive Probabilistic Hough Transform (PPHT). At the planimetric level, a cadastral map issue from an urban database is employed to segment the cloud of planar clusters into residential facade portion. Finally, each portion is vertically delimited using heuristic approaches. The adapted PPHT allows the automatic modeling of facade outlines with a fine facade line detection and a low computation time. The adapted approach has been tested on a set of point clouds acquired in the city of Paris under real conditions. Examples and experimental results show the efficiency of the proposed approach.