Extracting polygonal building footprints from digital surface models: A fully-automatic global optimization framework

Mathieu Brédif (1) , Olivier Tournaire (1) , Bruno Vallet (1) , Nicolas Champion (1)

1 MATIS - Méthodes d'Analyses pour le Traitement d'Images et la Stéréorestitution

This paper presents a fully automatic framework to extract building footprints from a Digital Surface Model (DSM). The proposed approach may be decomposed in two steps, each of them relying on a global optimization solver. The first step aims to extract rectangular building footprints directly from the DSM using a Marked Point Process (MPP) of rectangles. We introduce an energy that prevents overlapping rectangles and aligns rectangle edges with DSM discontinuities. This energy is then embedded in a RJMCMC sampler coupled with a simulated annealing to find its global optimum. Then, the second step of our framework refines these extracted rectangles into polygonal building footprints. We first create an arrangement of line segments supporting the rectangle edges. The dual graph of this arrangement is then considered in a maximum flow optimization scheme to remove edges in the arrangement which do not correspond to building edges in the DSM. Finally, 3D results illustrate a fully automatic process to build a 3D city model from a DSM only.

Mots clés: Urban; Building Modeling; Geometry Analysis; DEM/DTM

Domaines: Traitement des images [eess.IV]

Mathieu Brédif, Olivier Tournaire, Bruno Vallet, Nicolas Champion. Extracting polygonal building footprints from digital surface models: A fully-automatic global optimization framework. ISPRS Journal of Photogrammetry and Remote Sensing, 2013, 77, pp.57-65. ⟨10.1016/j.isprsjprs.2012.11.007⟩. ⟨hal-01557735⟩