TY - JOUR
T1 - Free multi-floor indoor space extraction from complex 3D building models
AU - Xiong, Qing
AU - Zhu, Qing
AU - Du, Zhiqiang
AU - Zlatanova, Sisi
AU - Zhang, Yeting
AU - Zhou, Yan
AU - Li, Yun
N1 - Accepted Author Manuscript
PY - 2016/11/25
Y1 - 2016/11/25
N2 - Intelligent navigation and facility management in complex indoor environments are issues at the forefront of geospatial information science. Indoor spaces with fine geometric and semantic descriptions provide a solid foundation for various indoor applications, but it is difficult to comprehensively extract free multi-floor indoor spaces from complex three-dimensional building models, such as those described using CityGML LoD4, with existing methods for the subdivision or extraction of indoor spaces based on vector topology processing. Therefore, this paper elaborates a new voxelbased approach for extracting free multi-floor indoor spaces from 3D building models. It transforms the complicated vector processing tasks into a simple raster process that consists of three steps: voxelization with semantic enhancement, voxel classification, and boundary extraction. Experiments illustrate that the proposed method can automatically and correctly extract free multi-floor indoor spaces, especially two typical kinds of open indoor spaces, namely, lobbies and staircases.
AB - Intelligent navigation and facility management in complex indoor environments are issues at the forefront of geospatial information science. Indoor spaces with fine geometric and semantic descriptions provide a solid foundation for various indoor applications, but it is difficult to comprehensively extract free multi-floor indoor spaces from complex three-dimensional building models, such as those described using CityGML LoD4, with existing methods for the subdivision or extraction of indoor spaces based on vector topology processing. Therefore, this paper elaborates a new voxelbased approach for extracting free multi-floor indoor spaces from 3D building models. It transforms the complicated vector processing tasks into a simple raster process that consists of three steps: voxelization with semantic enhancement, voxel classification, and boundary extraction. Experiments illustrate that the proposed method can automatically and correctly extract free multi-floor indoor spaces, especially two typical kinds of open indoor spaces, namely, lobbies and staircases.
KW - Free multi-floor indoor space
KW - CityGML LoD4
KW - Indoor space extraction
KW - Voxel
UR - http://resolver.tudelft.nl/uuid:76878988-c401-4262-bc56-35282c95e298
U2 - 10.1007/s12145-016-0279-x
DO - 10.1007/s12145-016-0279-x
M3 - Article
SN - 1865-0481
VL - 10
SP - 69
EP - 83
JO - Earth Science Informatics
JF - Earth Science Informatics
IS - 1
ER -