Scribd
Upload
56 views2 pages

Kecheng ISMLG

This document summarizes a research paper that proposes a framework using phase-based motion magnification and optical flow to monitor tunnel deformation during construction. The framework extracts the deformation signal from video frames taken inside a tunnel. It is able to match the temporal convergence trend and spatial deformation shape estimated from traditional automatic total station monitoring. The average error in convergence estimation using the proposed method is 9.04%. The framework could facilitate understanding tunnel-ground interaction and risk mitigation during construction.

Uploaded by

Luo Zhi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
56 views2 pages

Kecheng ISMLG

This document summarizes a research paper that proposes a framework using phase-based motion magnification and optical flow to monitor tunnel deformation during construction. The framework extracts the deformation signal from video frames taken inside a tunnel. It is able to match the temporal convergence trend and spatial deformation shape estimated from traditional automatic total station monitoring. The average error in convergence estimation using the proposed method is 9.04%. The framework could facilitate understanding tunnel-ground interaction and risk mitigation during construction.

Uploaded by

Luo Zhi
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

ISMLG 2023

4th International Symposium on Machine Learning and Big Data in Geoscience


University College Cork, 29th August – 1st September 2023

Deformation Monitoring of Tunnel using Phase-based Motion


Magnification and Optical Flow
Kecheng Chen*1, Hiroshi Kogi2 and Kenichi Soga1
1Civil and Environmental Engineering Department, UC Berkeley, United States
2Civil Engineering Technology Division, Shimizu Corporation, Japan
*presenting author (email: [email protected])

Keywords: Tunnel monitoring, Motion magnification, Optical flow

1 INTRODUCTION
During construction, continuous monitoring of underground tunnels can facilitate an in-depth
understanding of the ground-tunnel interaction and mitigate potential hazards [1]. Traditional
vision-based monitoring can directly capture a large range of motion but cannot separate the
tunnel’s vibration and deformation mode [2]. Phase-based motion magnification (PMM) is a
technique to magnify the motion in target frequency bands [3]. But most research related to
PMM has focused exclusively on surface structures [4]. Optical flow (OF) is a method for
motion calculation and has a much lower computational cost than Digital Image Correlation
(DIC) [5]. This paper proposes a PMM-OF-based monitoring framework to quantify the
deformation-induced displacements of the underground tunnel. The framework is used to
investigate the behavior of an existing tunnel, RMT2, when the nearby underground station
was being constructed [2]. R27, shown in Figure 1, was the instrumented RMT2 ring whose
center was directly above the tunneling center line. The prism at the crown is S4P3, and the
one at the right axis is S4P4. An automatic total station (ATS) was installed to monitor prisms
on the wall, and a Canon DSLR 600D camera was installed to monitor the whole tunnel.

2 DEFORMATION MONITORING FRAMEWORK


The framework can be explained as the following. First, high-resolution image sequences are
downscaled and converted to grayscale to avoid the high computational overhead. Second, the
low-pass Gaussian filter is used to correct the non-uniform illumination in the tunnel scenario.
Third, the complex steerable pyramid in PMM decomposes images into magnitude and phase
components. With temporal filtering, phase components in long change periods (>12h) are
increased by multiplying 1+α, where α is the magnification factor (usually 15). Fourth, the
original magnitude and renewed phase components are combined to reconstruct frames in the
deformation mode. Then the 2D Wiener filter is used to smooth the artifacts caused by the
large motion. Next, the GPU-accelerated deep-learning dense OF algorithm FlowNet2 is used
to estimate the absolute motion of each pixel [5]. Finally, the relative motion among pixels at
each tunnel ring is evaluated with the spatiotemporal median filter and scaled based on the
scaling factor to generate the deformation map. The scaling factor for each tunnel ring can be
calculated assuming the dimension of the prisms is known.
Kecheng Chen, Hiroshi Kogi and Kenichi Soga

3 RESULTS & DISCUSSIONS


The comparison of the convergence estimation between PMM-OF and ATS (benchmark)
monitoring is shown in Figure 2, where the two trends can match well. The average
convergence estimation error of PMM-OF monitoring is 9.04%, caused by the stacking of
camera tilting and motion magnification artifacts. Six points are sampled along R27 to
generate the deformation shape through PMM-OF monitoring, as shown in Figure 3(b), which
can well match the shape estimated by Alhaddad [2] through the combination of ATS
monitoring and Sattar Image Tracking (SIT), as shown in Figure 3(a).

Figure 1:
Instrumented area
(a) (b)
(The red area shows
R27, and the blue Figure 2: Comparison between Figure 3: Visualization of the
area shows the PMM-OF and ATS monitoring tunnel ring deformation on
monitored prisms) from 11/21/2014 to 12/01/2014 12/01/2014, where (a) is
(The convergence between the estimated by Alhaddad [2], and
S4P3 and S4P4) (b) is estimated by PMM-OF

4 CONCLUSIONS
This paper presents a PMM-OF-based deformation monitoring framework and estimates the
deformation of an existing tunnel disturbed by the nearby construction. The results indicate
that the temporal deformation trend and the spatial deformation shape estimated from PMM-
OF monitoring can well match that estimated from ATS monitoring.

REFERENCES
[1] Gue, C. Y., Wilcock, M., Alhaddad, M. M., Elshafie, M. Z. E. B., Soga, K., & Mair, R. J.
(2015). The monitoring of an existing cast iron tunnel with distributed fibre optic sensing
(DFOS). Journal of Civil Structural Health Monitoring, 5(5), 573-586.
[2] Alhaddad, M. (2016). Photogrammetric monitoring of cast‐iron tunnels and applicabilty of
empirical methods for damage assessment. University of Cambridge.
[3] Wadhwa, N., Rubinstein, M., Durand, F., & Freeman, W. T. (2013). Phase-based video
motion processing. ACM Transactions on Graphics (TOG), 32(4), 1-10.
[4] Fioriti, V., Roselli, I., Tatì, A., Romano, R., & De Canio, G. (2018). Motion Magnification
Analysis for structural monitoring of ancient constructions. Measurement, 129, 375-380.
[5] Ilg, E., Mayer, N., Saikia, T., Keuper, M., Dosovitskiy, A., & Brox, T. (2017). Flownet 2.0:
Evolution of optical flow estimation with deep networks. In Proceedings of the IEEE
conference on computer vision and pattern recognition (pp. 2462-2470).

You might also like