The full-field displacement intelligent measurement of retaining structures using UAV and 3D reconstruction

•A novel displacement measurement method for retaining structures is proposed.•We improve the original 3D reconstruction algorithm and propose the Improved-PatchmatchNet algorithm with higher training efficiency and computational efficiency.•An improved ICP algorithm with selectable alignment region...

Full description

Saved in:
Bibliographic Details
Published in:Measurement : journal of the International Measurement Confederation 2024-03, Vol.227, p.114311, Article 114311
Main Authors: Zhou, Xulin, Zhu, Qiankun, Zhang, Qiong, Du, Yongfeng
Format: Article
Language:English
Subjects:
3D reconstruction
Deep Learning
full-field Displacement
M3C2
PBGM
Retaining wall
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A novel displacement measurement method for retaining structures is proposed.•We improve the original 3D reconstruction algorithm and propose the Improved-PatchmatchNet algorithm with higher training efficiency and computational efficiency.•An improved ICP algorithm with selectable alignment region is proposed.•The PBGM algorithm evaluation framework was employed to evaluate the accuracy of algorithms.•This method can get the full-field displacement of the retaining structure. The displacement of retaining structures can provide an analytical basis for assessing its working condition. To address the shortcomings of conventional measurement techniques, which often involve high risk and labor-intensive procedures. This paper proposes an innovative approach. This method proposes the Improved-PatchmatchNet algorithm, Improved-ICP algorithm, and introduces the comparison algorithm M3C2 to achieve full-field displacement intelligent measurement of retaining structures. It is convenient, cost-effective, and acceptably accurate. In order to establish the practicality of this method, a physics-based graphic model (PBGM) is employed. During the PBGM model tests, it is observed that the error of the algorithm is less than 1 cm and the error of the average value is less than 6.82 %, which proved its effectiveness to measure the displacement of retaining structures. Ultimately, this method was employed to measure a framed composite anchored soil nail wall. Through six UAV flights over 18 months, five displacement measurements are obtained.
ISSN:0263-2241
1873-412X
DOI:10.1016/j.measurement.2024.114311