| spelling |
oapen-20.500.12657-891312024-04-03T02:23:56Z Chapter Quantifying the Confidence in Models Outputted by Scan-To-BIM Processes Bueno Esposito, Martin Malihi, Shirin Bosche, Frederic BIM point cloud confidence indoor modelling wall digital twin thema EDItEUR::U Computing and Information Technology::UT Computer networking and communications::UTV Virtualization 3D spatial data is increasingly employed to generate Building Information Models (BIMs) by extension digital twins for various applications in the architecture, engineering, and construction (AEC) sector such as project monitoring, engineering analyses, retrofit planning, etc. The outputted models of Scan-to-BIM processes should satisfy pre-defined levels of quality. In the case of emerging automated Scan-to-BIM solutions, users however currently need to check all generated geometry manually, which is time-consuming. What would help users is if the automated systems could also provide a level of confidence in the detection and modelling of each element. In this paper three generic indicators are defined for analysing the reliability of the generated 3D models: Icoverage estimates the portion of the surface of the modelled element that can be explained by the input point cloud. Idistance defines the closeness of the generated element models to the input point cloud. The confidence of the generated 3D local models can be computed by combining the two aforementioned indices. The proposed indicators are assessed using actual examples and comparisons are conducted between automatically generated 3D BIM models and 3D models generated manually by a BIM modeler 2024-04-02T15:47:28Z 2024-04-02T15:47:28Z 2023 chapter ONIX_20240402_9791221502893_100 2704-5846 9791221502893 https://library.oapen.org/handle/20.500.12657/89131 eng Proceedings e report application/pdf n/a 9791221502893_113.pdf https://books.fupress.com/doi/capitoli/979-12-215-0289-3_113 Firenze University Press 10.36253/979-12-215-0289-3.113 10.36253/979-12-215-0289-3.113 bf65d21a-78e5-4ba2-983a-dbfa90962870 9791221502893 137 10 Florence open access
|
| description |
3D spatial data is increasingly employed to generate Building Information Models (BIMs) by extension digital twins for various applications in the architecture, engineering, and construction (AEC) sector such as project monitoring, engineering analyses, retrofit planning, etc. The outputted models of Scan-to-BIM processes should satisfy pre-defined levels of quality. In the case of emerging automated Scan-to-BIM solutions, users however currently need to check all generated geometry manually, which is time-consuming. What would help users is if the automated systems could also provide a level of confidence in the detection and modelling of each element. In this paper three generic indicators are defined for analysing the reliability of the generated 3D models: Icoverage estimates the portion of the surface of the modelled element that can be explained by the input point cloud. Idistance defines the closeness of the generated element models to the input point cloud. The confidence of the generated 3D local models can be computed by combining the two aforementioned indices. The proposed indicators are assessed using actual examples and comparisons are conducted between automatically generated 3D BIM models and 3D models generated manually by a BIM modeler
|
