Indoor mapping and modeling by parsing floor plan images

ABSTRACT

A large proportion of indoor spatial data is generated by parsing floor plans. However, a mature and automatic solution for generating high-quality building elements (e.g., walls and doors) and space partitions (e.g., rooms) is still lacking. In this study, we present a two-stage approach to indoor mapping and modeling (IMM) from floor plan images. The first stage vectorizes the building elements on the floor plan images and the second stage repairs the topological inconsistencies between the building elements, separates indoor spaces, and generates indoor maps and models. To reduce the shape complexity of indoor boundary elements, i.e., walls and openings, we harness the regularity of the boundary elements and extract them as rectangles in the first stage. Furthermore, to resolve the overlaps and gaps of the vectorized results, we propose an optimization model that adjusts the rectangle vertex coordinates to conform to the topological constraints. Experiments demonstrate that our approach achieves a considerable improvement in room detection without conforming to Manhattan World Assumption. Our approach also outputs instance-separate walls with consistent topology, which enables direct modeling into Industry Foundation Classes (IFC) or City Geography Markup Language (CityGML).

KEYWORDS:

• Floor plan
• image vectorization
• topological consistency
• indoor mapping and modeling
• indoor location-based services

Acknowledgments

This study is supported by the National Key Research and Development Program of China (No.2016YFB0502203) and the National Natural Science Foundation of China (No. 41271440). Moreover, we are also very grateful to Professsor Xiaohui Yuan at University of North Texas for the useful discussions. Last but not least, we thank Xin Li, Ruqin Zhou, and Decheng Cai, post-graduates of Ubiloc team at China University of Geosciences, Wuhan, for their contributions to the experiments.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data and codes availability statement

Our experiments are conducted on the public dataset CVC-FP released at http://dag.cvc.uab.es/resources/floorplans/. Meanwhile, our codes, processed datasets for 5-fold cross-validation, and the corresponding results are released with a DOI as https://doi.org/10.6084/m9.figshare.12082131.

Notes

1. We used the PyTorch version code released by the authors at https://github.com/art-programmer/FloorplanTransformation and trained on the CVC-FP dataset to generate this result.

Additional information

Funding

This work was supported by the National Key R&D Program of China [2016YFB0502203]; National Natural Science Foundation of China [41271440].

Notes on contributors

Yijie Wu

Yijie Wu received her Master degree in Cartography and Geographic Information Engineering from the School of Geography and Information Engineering, China University of Geosciences, Wuhan. Her research interests are indoor mapping and modeling.

Jianga Shang

Jianga Shang is currently a Professor with the School of Geography and Information Engineering, China University of Geosciences, Wuhan. He received the Ph.D. degree in computer system architecture from the Huazhong University of Science and Technology. He is also the director of research group for intelligent system Software technology in the National Engineering Research Center for Geographic Information System. Dr Shang is a committee member of ISPRS WG IV/5 on indoor/outdoor seamless modeling, LBS, and mobility. He is also a member of the IEEE, ACM, China Computer Federation (CCF), and GNSS & LBS Association of China(GLAC). His research interests include indoor location-awareness, mobile and context-aware computing, geospatial information systems, and human-cyber-physical integration software.

Pan Chen

Pan Chen is a lecturer in the School of Geography and Information Engineering, China University of Geosciences, Wuhan. He received the Ph.D. degree in Huazhong University of Science and Technology (HUST), Wuhan, China, in 2016. His current research areas include indoor localization and pattern recognition. He is a member of the China Computer Federation.

Sisi Zlatanova

Sisi Zlatanova is currently a Professor and the Head of GRID, Faculty of the Built Environment, The University of New South Wales, Sydney, Australia. Her research interests are in modeling, management, and analysis of 3D built environment.

Xuke Hu

Xuke Hu is a Ph.D. candidate at the GIScience group of Heidelberg University since 2015. He is now a full-time researcher at the data science institute of the German Aerospace Center (DLR). His research interests include indoor localization/navigation/mapping, VGI, social media data analysis, earth observation, and disaster management.

Zhiyong Zhou

Zhiyong Zhou is a Ph.D. student of Geographic Information Systems unit in the Department of Geography, University of Zurich, Switzerland. His research interests include location-based services (LBS), computational place modeling, and wayfinding.