Investigating the accuracy of 3D models created using SfM

Research output: Contribution to conferencePaper

Abstract

Technological innovations in photogrammetry have progressed over time, and civil engineers in Japan have employed this technology for earthworks. Photogrammetry has also been introduced to building construction work, and in particular, it is used in the surveying of buildings and the production of architectural drawings in repair work. With this motivation, the authors evaluate the accuracy of Three-dimensional (3D) model extraction using structure from motion (SfM) photogrammetry. In particular, the authors evaluate the accuracy of the height measurements (Z axis). The authors survey the accuracy of a 3D model using a cube with a side length of 5 cm. Our surveying procedure consists of the following four steps: 1. Place the cube on a turntable, rotate the turntable in 5° increment, and capture 72 photographs. 2. Select 12 photographs considering six control points. Edit the control points given their 3D positions in each of the 12 photographs. 3. Extract the 3D model using SfM software. 4. Import the 3D model and measure the height of the cube at 121 points. In this paper, we discuss the differences in the measurement accuracy resulting from the differences in the camera position.

Original languageEnglish
Pages834-839
Number of pages6
Publication statusPublished - 2017 Jan 1
Externally publishedYes
Event34th International Symposium on Automation and Robotics in Construction, ISARC 2017 - Taipei, Taiwan, Province of China
Duration: 2017 Jun 282017 Jul 1

Other

Other34th International Symposium on Automation and Robotics in Construction, ISARC 2017
CountryTaiwan, Province of China
CityTaipei
Period17/6/2817/7/1

Keywords

  • Accuracy
  • Height
  • Photogrammetry
  • SfM

ASJC Scopus subject areas

  • Building and Construction
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Investigating the accuracy of 3D models created using SfM'. Together they form a unique fingerprint.

  • Cite this

    Ishida, K. (2017). Investigating the accuracy of 3D models created using SfM. 834-839. Paper presented at 34th International Symposium on Automation and Robotics in Construction, ISARC 2017, Taipei, Taiwan, Province of China.