Three-dimensional (3D) reconstruction of a building from terrestrial laser scanning and photogrammetry dataset using Identical Point Picking (IPP) registration method

Building is an immovable asset that requires accurate mapping and efficient documentation. Limitation of 2D- based data acquisition and insufficient semantic information on its geometry make the building documentation becomes less effective. Plus, even 3D- based data acquisition is used, it does not...

Full description

Saved in:
Bibliographic Details
Main Author: Razali, Ahmad Firdaus
Format: Thesis
Language:English
Published: 2022
Subjects:
Online Access:http://eprints.utm.my/id/eprint/100138/1/AhmadFirdausRazaliMFABU2022.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Building is an immovable asset that requires accurate mapping and efficient documentation. Limitation of 2D- based data acquisition and insufficient semantic information on its geometry make the building documentation becomes less effective. Plus, even 3D- based data acquisition is used, it does not cover whole details optimally. The objectives of this research are to combine two geospatial surveying techniques; terrestrial laser scanning (TLS) and unmanned aerial vehicle (UAV) photogrammetry dataset into 3D modelling and to determine accuracy of the point clouds and the rendered 3D model. A geodetic laser scanner-2000 (GLS- 2000) terrestrial laser scanner was used to record existing details of Dewan Muafakat Johor, Taman Kobena, Tampoi, Johor Bahru, Johor. Meanwhile, UAV-photogrammetry was utilized to capture the rooftop and the aerial view of the building. Both pointcloud data from TLS and UAV-photogrammetry are integrated using Identical Point Picking (IPP) registration method until combined clouds called hybrid point cloud dataset are produced. A level of detail (LOD 3) of a three-dimensional (3D) model was developed using Sketchup and Undet For Sketchup (UFS) software based on the hybrid pointcloud. The root mean square error (RMSE) obtained shows that point cloud dataset is more accurate with 0.132m compared to higher RMSE, 0.455m of rendered 3D model dataset. As this study focuses on 3D data representation which significant for asset documentation, building monitoring and building information modelling (BIM) are the example of applications in the industry that suit with the purpose. Hence, the output from this study especially in terms of measurement and accuracy would be referable for decision making in building documentation.