A new reconstruction method for 3D buildings from 2D vector floor plan - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2013

A new reconstruction method for 3D buildings from 2D vector floor plan

Junfang Zhu
  • Function : Author
Hui Zhang
  • Function : Author
Yamei Wen
  • Function : Author


This paper proposes a method to analyze the geometry and semantic information of2D vector floor plans, and reconstruct the corresponding 3D building modelsautomatically. First, the Shape-opening graph (SOG) is introduced to recognizeStructural Components (SCs) and describe the relationships between SCs and openingswhich are architectural components separating spaces. A priority principle algorithmis developed to replace openings with wall equivalent lines for the purpose of laterloop searching. Then, we present an odd-even-based edge breaking algorithm topreprocess wall lines in order to search loops directly and exactly. After that, all theinterior-space contours, which are in fact rooms or aisles, and exterior boundarycontour can be obtained. With the hierarchical component tree presented, thetopological relations of the whole building can be obtained. Finally, 3D models can begenerated after extruding loops to the wall height and cutting openings from the walls.A drawing with several floor plans can also be reconstructed to multi-floor buildingafter aligning their dimensions. Moreover, the intermediate result after analyzing the2D floor plan can be the input of other architectural software to generate 3D models.
Fichier principal
Vignette du fichier
ae_a_e_2013a_3De_e_a_.pdf (277.4 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-00920680 , version 1 (19-12-2013)


  • HAL Id : hal-00920680 , version 1


Junfang Zhu, Hui Zhang, Yamei Wen. A new reconstruction method for 3D buildings from 2D vector floor plan. Computer-Aided Design and Applications 2013, Jun 2013, Lombary, Italy. ⟨hal-00920680⟩
241 View
2826 Download


Gmail Facebook X LinkedIn More