Next Article in Journal
Satellite Image Super-Resolution via Multi-Scale Residual Deep Neural Network
Next Article in Special Issue
HBIM for Conservation: A New Proposal for Information Modeling
Previous Article in Journal
RFim: A Real-Time Inundation Extent Model for Large Floodplains Based on Remote Sensing Big Data and Water Level Observations
Previous Article in Special Issue
Classification of 3D Digital Heritage
Open AccessArticle

Clustering of Wall Geometry from Unstructured Point Clouds Using Conditional Random Fields

Department of Civil Engineering, TC Construction—Geomatics, KU Leuven—Faculty of Engineering Technology, 9000 Ghent, Belgium
Author to whom correspondence should be addressed.
Remote Sens. 2019, 11(13), 1586;
Received: 14 May 2019 / Revised: 11 June 2019 / Accepted: 12 June 2019 / Published: 4 July 2019
(This article belongs to the Special Issue Heritage 3D Modeling from Remote Sensing Data)
The automated reconstruction of Building Information Modeling (BIM) objects from point cloud data is still subject of ongoing research. A vital step in the process is identifying the observations for each wall object. Given a set of segmented and classified point clouds, the labeled segments should be clustered according to their respective objects. The current processes to perform this task are sensitive to noise, occlusions, and the associativity between faces of neighboring objects. The proper retrieval of the observed geometry is especially important for wall geometry as it forms the basis for further BIM reconstruction. In this work, a method is presented to automatically group wall segments derived from point clouds according to the proper walls of a building. More specifically, a Conditional Random Field is employed that evaluates the context of each wall segment in order to determine which wall it belongs to. First, a set of classified planar primitives is obtained through algorithms developed in prior work. Next, both local and contextual features are extracted based on the nearest neighbors and a number of seeds that are heuristically determined. The final wall clusters are then computed by decoding the graph. The method is tested on our own data as well as the 2D-3D-Semantics (2D-3D-S) benchmark data of Stanford. Compared to a conventional region growing method, the proposed method reduces the rate of false positives, resulting in better wall clusters. Overall, the method computes a more balanced clustering of the observations. A key advantage of the proposed method is its capability to deal with wall geometry in complex configurations in multi-storey buildings opposed to the presented methods in current literature. View Full-Text
Keywords: building information modeling; 3D modeling; semantic classification; point clouds building information modeling; 3D modeling; semantic classification; point clouds
Show Figures

Figure 1

MDPI and ACS Style

Bassier, M.; Vergauwen, M. Clustering of Wall Geometry from Unstructured Point Clouds Using Conditional Random Fields. Remote Sens. 2019, 11, 1586.

Show more citation formats Show less citations formats
Note that from the first issue of 2016, MDPI journals use article numbers instead of page numbers. See further details here.

Article Access Map by Country/Region

Search more from Scilit
Back to TopTop