Point cloud data is a collection of points in a three-dimensional (3D) space. Measurements of these points can include the x, y, z coordinates, vector information, colour, and other relevant information related to the points. These points can be joined together to form objects or surfaces, which can then be used to build a complete model of the building.

To collect the point cloud data, two different systems may be deployed. A laser scanning system may be used to generate the data. In this system, a laser beam is transmitted and bounces off a specific point in the room, and the time taken for the signal to return to the original location is measured. Using the time taken, the distance from the equipment to the target surface is measured. This is done repeatedly to generate a point cloud for the whole room.

Another system to collect point cloud is photogrammetry, where multiple views of the same location are photographed from different angles. An object located in the same location in the room may appear slightly different due to the different angles. By using the information collected, an algorithm is applied to reconstruct a 3D model from two-dimensional (2D) images.

Table 1. Advantages and Disadvantages of Each System

Applications for construction and building usage requires models to have high accuracy with the measurements. As such, laser scanning is generally favoured as the data collection method for 3D point cloud data in the construction industry. There are many applications for the point cloud data collected.
Automatic Creation of As-Built BIMs

Given a point cloud of a facility, the modelling of BIM involves three tasks: 1) modelling the geometry of the components, 2) assigning an object category and material properties to a component, and 3) establishing relationships between the components. For example, when modelling a wall; firstly, we need to establish the shape of the wall; secondly, categorise the model created as a brick wall; and lastly, connect Wall1 to Wall 2 within the specific location. The as-built BIM creation process covers the three core operations: geometric modelling, object recognition, and relationship modelling.

Geometric modelling is the process of constructing simplified representations of the 3D shape of building components, such as walls, windows, and doors, from point cloud data. The output representation can be either parametric or nonparametric, as well as surface-based or volumetric. Here, we focus on parametric surface modelling methods. These are most relevant to the construction of BIMs. Parametric surface modelling requires the detection and extraction of geometric primitives and their parameters. Common types of primitives occurring in facilities include planar surfaces, curved surfaces (such as cylinders and cones), and extrusions (such as decorative mouldings and trim). The second core task of as-built BIM construction is object recognition, the process of labelling a set of data points or geometric primitives extracted from the data with a named object or object class. For example, the modelling task would find a set of points to be a vertical plane, the recognition task would label that plane as being a wall. Object recognition algorithms may label object instances of an exact shape, or they may recognize classes of objects, where the shape may vary among instances from the class (e.g. recognize all windows that can vary in height).

Spatial relationships between objects in a BIM are useful in many scenarios. So, common relationships modelled in a building information model include aggregation relationships (e.g. a window is contained in a wall), topological relationships (e.g. Wall1 is connected to Wall2), and directional relationships (e.g. the second floor is above the first floor). Several spatial relationship models have been developed for automatically deriving topological relationships between objects. For example, some researchers use a 3D solid CAD model to automatically derive topological relationships between solid objects or geometric primitives.

Related literature indicates a plethora of applications being made using 3D point cloud data (Ma and Liu, 2018). One common application includes construction progress tracking in steel or concrete structures, earthwork, fake structures simulating columns, and in secondary or temporary objects. Building renovation activities are also supplemented with point cloud data for re-design and renovation of façade, interior design, refurbishments, and energy performance retrofit. On the other hand, applications such as digital reproduction, automated earthwork excavation and robot navigation (in indoor and outdoor built environments) showcase how this technology assists in construction automation. Further applications include building performance analysis, construction safety management and progress management. Where point clouds are used in finite element modelling of mechanical or structural analysis, morphologic analysis, stress estimation, accessibility diagnosis, energy performance modelling, blind spot identification (e.g. of construction equipment), safety hazard identification and simulation, active safety management, and safety assistance for mobile crane lifting operations.

A much more common application of 3D point clouds is the heritage application. These include surveying, repair and maintenance, stability analysis, documentation, and site preservation of heritage buildings. During these applications, a 3D reconstruction of historic buildings is being made. The virtual modelling and reconstruction of heritage buildings are done for conservation purposes using a combination of graphical and semantic information from point clouds and bibliographical data (López et al., 2017). These reconstructions of heritage buildings are useful for digitally replacing physical (e.g. damaged or missing) artefacts, research and educational orientation, entertainment (e.g. heritage building representation in games), and for business purposes. The reason for this technology to be used in such a wide range of applications is owed to safe and faster implementation times, ability to reconstruct structures accurately and completely, less intrusive, more informative, and lower cost as compared to other alternatives.

References

• “3D Photogrammetry VS 3D Laser Scanning, Who Wins?” AVRspot, 7 Aug. 2019, www.avrspot.com/3d-photogrammetry-3d-laser-scanning/.
• Leica Geosystems AG. Laser Scanning: Chapter 2 of 3 - How It All Works. Youtube, www.youtube.com/watch?v=1lDO1UevAJI.
• “Making 3D Models with Photogrammetry.” The Haskins Society - 3D Photogrammetry with PhotoScan, thehaskinssociety.wildapricot.org/photogrammetry.
• Tang, P., Huber, D., Akinci, B., Lipman, R. and Lytle, A., 2010. Automatic reconstruction of as-built building information models from laser-scanned point clouds: A review of related techniques. Automation in construction, 19(7), pp.829-843.
• López, F.J., Lerones, P.M., Llamas, J., Gómez-García-Bermejo, J. and Zalama, E., 2017. A framework for using point cloud data of heritage buildings toward geometry modeling in a BIM context: A case study on Santa Maria La Real De Mave Church. International Journal of Architectural Heritage, 11(7), pp.965-986.
• Ma, Z. and Liu, S., 2018. A review of 3D reconstruction techniques in civil engineering and their applications. Advanced Engineering Informatics, 37, pp.163-174.