Automated Floor Plan and Building Model Creation for Cultural Heritage Buildings from Laser Scanner Data
Date
2021-06-21
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
This research project developed and implemented an automated modelling system to create 2D floor plans and 3D building models of heritage sites. Without building plans, it is more difficult for an historic building to receive historic designation and restoration funding. Under current practice, the creation of such physical documentation is expensive and time-consuming. Physical documentation can include as-built architectural plans, elevations, profiles and photographs, whereas historic documentation includes important documents, artefacts, and historic photographs or archives. Important heritage sites whose building plans have been lost or destroyed or become inaccurate through renovations are often left abandoned or not kept up properly because they are unable to receive the necessary support. The current modelling process involves the utilization of CAD software and a trained modeller to digitally draw a 2D floor plan, or a more complex 3D building model, overlain upon the point cloud data collected by a laser scanner. As currently applied, point cloud modelling requires inefficient manual manipulation, editing and rendering of large datasets within a CAD environment to produce floor plans and building models. This research project used statistical methods such as principal components analysis (PCA) and M-estimator sample consensus (MSAC) to automatically detect building features from a point cloud captured through a 3D terrestrial laser scan (TLS) of the building site. Two novel methods were developed in this work to help in the automation of floor plan and building model creation. The first was a novel methodology for the automated separation of storeys within a multi-level, multi-storey building. The second was a novel methodology for the automated detection of doors and windows within a point cloud using a wall-defined search space. These new methods were implemented as components in an end-to-end modelling strategy for the creation of floor plans and building models, the final output of which is written in a CAD-accessible file format. The modelling strategy showed an overall accuracy of 92.75% for the tested datasets, demonstrating the ability of the developed program to accurately produce both 2D floor plans and 3D building models of multi-level building storeys with door and window features. The development of this automated process will allow a non-geomatics expert to create floor plans and/or building models of sites with significantly less manual effort and reduced cost. This will increase the ability of heritage sites to receive historic designation, allowing them to be better preserved over time.
Description
Keywords
Terrestrial Laser Scanning, Digital Heritage, Building Models, Floor Plans, Point Cloud Processing
Citation
Pexman, K. (2021). Automated Floor Plan and Building Model Creation for Cultural Heritage Buildings from Laser Scanner Data (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca.