Abstract

This thesis focuses on the problems surrounding Structure from Motion (SfM) data capturing and analysis in the context of 3D surface inspection. The use of 3D point clouds and meshes has become more widely spread, because of the additional level of information that they contain compared to more traditionally used inspection data like 2D images.
Through the thesis, each part of the SfM pipeline is examined. From looking at best practices when capturing images both terrestrial and in the air using unmanned aerial vehicles and developing simulation environments for testing capturing setups. Through the development of automatic methods for detecting the absolute scale of the reconstructed surfaces, using additional information from GPS and LiDAR data. Finally, analyzing the resultant reconstructions to determine their accuracy, the presence of noise, and possible use cases, like conveying the surface profile through vibrations and segmenting it into different roughness areas.
Original languageEnglish
PublisherAalborg Universitetsforlag
Number of pages292
ISBN (Electronic)978-87-7210-818-6
Publication statusPublished - 2021
SeriesPh.d.-serien for Det Tekniske Fakultet for IT og Design, Aalborg Universitet
ISSN2446-1628

Bibliographical note

PhD supervisor:
Professor Claus Madsen, Aalborg University

Keywords

  • Structure from Motion
  • 3D reconstruction
  • computer vision
  • surface analysis
  • metrology
  • virtual reality

Fingerprint

Dive into the research topics of 'Applied SfM - Capturing and Analysing 3D Surface Data'. Together they form a unique fingerprint.

Cite this