System Design of 3D Reconstruction of Rock Deformation in Tunnels Based on a Deep Convolutional Neural Network

This paper presents a high-precision 3D method for non-contact measurement of the deformation of tunnel surrounding rock. The method uses a 2D high-resolution photograph taken with a monocular camera and a 3D reconstruction technique to obtain a 3D image of tunnel surrounding rock. First, the authors analyze the basic principles of 3D-image-reconstruction technology. Next, the authors discuss methods for calculating the 3D shape of the known and unknown 2D key points and conclude that the use of deep convolutional neural network is an effective method and a new idea. Finally, the authors present findings that demonstrate that the 3D reconstruction method proposed in this paper can be used to obtain tri-dimensional models and deformation variables for tunnel surrounding rock that are superior to results obtained using traditional methods. The method is simple, provides good visibility, has a high degree of accuracy, and does not affect the physical structure of the tunnel.

Language

  • English

Media Info

  • Media Type: Web
  • Pagination: pp 1531-1545
  • Monograph Title: CICTP 2020: Advanced Transportation Technologies and Development-Enhancing Connections

Subject/Index Terms

Filing Info

  • Accession Number: 01749852
  • Record Type: Publication
  • ISBN: 9780784482933
  • Files: TRIS, ASCE
  • Created Date: Aug 12 2020 3:03PM