1.

INTRODUCTION

In recent years, the use of terrestrial laser scanning technique in engineering surveys is
gaining an increasing interest due to the advantages of non-contact, rapidity, high
accuracy and large scale. This technique delivers millions of accurate 3D points (mm
level deformation measurement; cross-section extraction; curacy) with a very high point
density in a short time (up to 1 million points per second) ，which makes it a valuable
alternative or complementary technique for classical topographical measurements based
on total station or digital photogrammetry． The terrestrial laser scanning can still deliver
very accurate points even in the situations where other topographical techniques are
difficult or impossible to use．

The digital photogrammetry is inapplicable under some extreme conditions, such as the
drilling tunnels, but the laser scanning is applicable in these complex situations． The
measurement with a total station is also an option, but the advantage of the laser scanning
is obvious: instead of focusing on the rather limited number of specified points, the laser
scanning delivers millions of 3D points in a complete monitored tunnel section．

Recently, the improvements of this technique regarding the speed, accuracy, software
algorithms， and the fall in price have introduced a high potential of large scale
applications in highly demanding engineering environments such as tunnels, bridges, and
heritage buildings. Tunnels in particular those with long lengths create great challenges
for surveyors due to difficulty to obtain the satisfactory geometry of the scanned data．

The high resolution point clouds provided by laser scanning techniques have several
applications in construction of tunnels such as construction survey of tunnels extraction
of cross-section or feature line of tunnels and deformation measurement of tunnels

2. TERRESTRIAL LASER SCANNING

2.1. LIDAR TECHNIQUES

The core technology of the terrestrial laser scanning is the LiDAR technique,
which is used to obtain the distance of each object point from the lens. The
acronym LiDAR stands for light detection and ranging. The laser system produces
and emits a beam (or a pulse series) of highly collimated，directional，coheren,

1
and in-phase electromagnetic radiation. When the light reflected by the surface of
an object is received the system can calculate the range by the flight time and
acquire the reflectivity of the surface Fig.2.1 shows one example of the laser
scanners.
There are two different methods of range determination: phase and pulse (
Jaboyedoff et al 2012). The former is more accurate in range but suffers from a
limited range. Alternatively, the latter can measure in a greater range. Therefore,
the latter is implemented in most TLS used for the measurement of civil
construction.

Figure 2.1 Example of laser scanner

2.2. MEASUREMENT PRINCIPLE
A laser scanner consists of a transmitter/receiver of the laser beams, a scanning
device and a timing device (Fig.2.2). The scanner sends out laser pulses then
receives and records the reflected signals. The timing device measures the time of
flight (Δt), with which the scanner can compute the distance d, where c stands for
light speed.
d = cΔt
Knowing the direction and the angle of the light (cos(α), cos(β), cos(γ)) allows
determining the relative position (x, y, z) of a reflective surface to the device．
x = dcos(α)
y = dcos(β)
z = dcos(γ)

2
 Figure 2.2 Principles of laser scanner data acquisition
2.3. ACCURACY, RESOLUTION, AND POINT DENSITY
The typical accuracy of a laser instrument is ±4 mm, within the maximum distance
of about 300 m. Nevertheless, the instrumental accuracy is usually lower in
practical applications due to unfavorable conditions such as bad weather
conditions (rain or fog) very low reflectivity of the object surfaces and very bright
ambient conditions, etc. The resolution of the laser scanner usually declines with
the distance due to the laser beam divergence. For example, the laser's spot
dimension of the HDS6100 (one type of laser scanner) increases 3 mm when the
distance increases 50 m.
Regardless of the increasing of the laser's spot dimension, the scanner records the
center of the spot as a point; therefore, the point density is lower than 1 mm even
in the maximum distance．

2.4. REGISTRATION OF POINT CLOUD

To obtain a 3D model of a construction the scanner must scan in different
positions. The several point clouds acquired from different positions require
registration to constitute the entire 3D model. The typical method is to place some
targets that can be recognized by the scanner before scanning. The scanner can
mark the targets as specific points and we can register different point clouds with
the targets they share

3
3. PROCEDURE FOR CAPTURING 3D DIGITAL DATA

A typical scanning procedure and the associated data can be described as the
following. For a typical scanning job in the field, both reference surveying and
scanning are needed.

3.1. CONTROL SURVEYING

Reference surveying is often carried out by using a total station to set up a local
coordinate system around the scanning area, and link the local co-ordinate system
to a global coordinate system, and then the surveying data are applied for
positioning of each scanning set into the global reference co-ordinate system.

3.2. SCANNING
The operation of each different scanning systems is different—because it depends
upon the requirements of scanning software and reference surveying. But some
basic parameters must be correctly selected for different applications: 1) Scanning
resolution; 2) Scanning range; 3) Position of each scanning; 4) Number and
location of reference targets. In addition, some scanning systems are also sensitive
to the environment, such as temperature, moisture, density of particles in the air,
and even the reflectivity of the object. These parameters and factors must be
carefully considered in order to obtain good quality scanning data.
The typical raw data from laser scanning is the so-called point cloud, which is a
set of vertices in a 3D co-ordinate system. These vertices are usually defined by
x-y-z coordinates, and typically intended to be digitally representative of the
external surface of the rock mass.
For most of the phase-based scanners, the raw scanning data is combined as both
point and intensity, so the corresponding intensity image in both 2D and 3D can
be obtained, which is useful for documentation and identification of objects in
detail. Figure 3.1 shows the raw scanning data typically captured from a phase-
based scanner.
The true colour scanner is still under development, and it is only possible now to
scan some small objects, but not possible in practice to scan the rock mass over a
large area. However, another alternative, through combining scanning data with
colour photos in software, is now possible. When scanning at the location, the
colour photos are captured by a digital camera at the same time, and then the

4
 colour photos are registered to the scanning data in the software; the true colour
3D model of a rock mass can be obtained, see Figure 3.2.

Figure 3.1 Presentation of raw scanning data from scanning of the same object

Figure 3.2 3D colour model of scanning in a tunnel

3.3. PRE-PROCESSING DATA
The on-site captured scanning data, also termed raw scanning data, are necessary
to make pre-processing before any further modelling and calculation. The reasons
are:
Each individual scan obtains the scanning data in the local co-ordinate system,
and mostly several scans are needed to cover the whole object. Therefore, each
individual scan needs to be registered into a common co-ordinate system.
Not all points in the raw scanning data can be used because there are always some
noise points captured, especially there are more noise points from the phase-based

5
scanner. So, these noise points need to be filtered out from the raw scanning data.
In this case, the raw scanning data need to be pre-processed mostly by two steps:
Step 1. Register each individual scan into a common co-ordinate system; and
Step 2. Filter out noise points from the raw scanning data.
In addition, there are different data formats for the raw scanning data from
different scanner producers, so it is sometimes necessary to convert the raw
scanning data into another format, depending upon which modelling software is
used for the post processing

3.4. POST-PROCESSING DATA

After pre-processing of the raw scanning data, the point clouds are positioned and
oriented in a certain co-ordinate system, and all scanning points are correctly
located in a known co-ordinate system, so the post-processing can be performed.
The post processing is done by two different situations or software, as follows.
1. By special developed software
As the scanning data are quite substantial, compared to other data, and also have
a special data format, special software has been developed. These software
methods can process large amounts of scanning data, and create different results,
including support for a CAD model, mesh-model, cross-section, etc. And then,
the results can be exported into other existing systems, such as CAD, GIS or other
user-familiar systems for different applications.
2. Input into existing software
This was difficult in the earlier period of the development, but now many types of
socalled ‘plug-in’ software have been developed, and this plug-in software makes
it possible to import a large scanning data file into user-familiar software, such as
CAD, GIS and so on.
In addition, many hardware producers will now attract more users, so they allow
their special data formats to be converted into a neutral format, the ascii format,
to be available for post-processing in many different software programs.

6
4. 3D MODELING

4.1. DATA ACQUISITION

4.1.1 Laser Scan Mode:
In the current system, laser data acquisition is done using a laser scanner and
manually distributed 3d targets for mutual orientation between adjacent scans.
Similar to the stereo photogrammetry mode, during laser scanning the 3d targets
are measured by a theodolite controlled via radio link by the scanning PC. This
allows fully automatic laser scanner orientation and on-line user feedback about
the orientation success. For high resolution texture data acquisition, within tunnels
the use of line-scan cameras for high resolution texture and/or to facilitate mutual
orientation is not easily applicable due to poor illumination conditions. Therefore,
high-resolution frame cameras using flashlight are used. Image acquisition and
preliminary image orientation in the current set-up can be done in two modes:
A camera is fixed on the sensor head and, together with the sensor, moves
around the main sensor axis (Figure 4.1, middle). In this case mutual
orientation between sensor axis and camera can be determined in the lab.
Proper synchronization between camera exposure and laser scanner is
required to determine the correct laser scanner angle during exposure.
Scanner and camera data are acquired seperately, an odometer on the
mobile platform enables rough knowledge about the camera poses within
the reference tunnel coordinate system (Figure 4.1, right).
In Option 2 the odometer has an accuracy of several tens of cm including a rough
initialization using well known positions within the tunnel. The position with
respect to the tunnel axis (elevation & horizontal offset) has the same order of
accuracy. The camera viewing angles are only known within a range of +/- 3
degrees, due to changing tilt of the vehicle. Option 1 does not lead to perfectly
aligned texture results either, since the interior orientation of consumer camera
optics is not stable enough when used in various rotational states. Moreover, the
mounting between laser scanner and camera is frequently separated, which
prevents a repeatable system calibration. Therefore, both acquisition modes
require a refinement of laser – to – camera orientation. This is accomplished by
matching between the laser reflectivity texture and the digital image.

7
 Figure 4.1 Data acquisition

4.2. DATA PROCESSING

Data acquisition provides either oriented stereo images, oriented laser scan point
clouds, or both. The first part of the data processing chain therefore differs
depending on the data source:
For the case of stereo images, stereo matching takes place followed by a
projection of the disparities onto the discrete tunnel model.
In case of laser scans the 3d co-ordinates and laser reflectivity values are
projected to the discrete tunnel model, optionally together with the
registered images.
Therefore, for each sensor data set the 3d reconstruction result is a high resolution
textured tunnel model segment represented as regular grid. The grid contains on
the one hand the deviations to the design model (“digital surface model image”),
on the other hand the camera (and/or laser reflectivity) image texture projected on
the design model (“ortho image”). The following steps can be done directly on
these images:
To refine the sensor orientations, mutual matching between adjacent
reconstructions is used for a global geometry – based adjustment. Finally the
texture mosaic is further refined by an automatic global non-linear stitching
process.

4.3. VISUALIZATION, PLOTTING AND MODELING

Data visualization takes place either in interactive mode (Module Analyse, see
Figure 4.2), or in the form of customer – defined plots. Analyse offers the ability
to display the colour ortho image in full resolution, real time display of profiles at
arbitrary locations, coordinates, distances in various modes, and a large set of
customer – defined statistical values such as volumes (within regions of interest,

8
 interpolated and/or extrapolated). Different construction phases (raw excavation,
shotcrete, inner lining etc.) can be visualized and overlaid simultaneously, which
allows a high- resolution registered comparison between them. To focus on
specific regions of interest or omit obsolete surface parts interactive masking is
possible. Plotting is realized making use of a commercial visualization suite which
handles the layout, colouring and annotation for profile plots and virtually infinite
texture plots, optionally overlaid with false colouring, iso-lines for radial
deviations from the design profile, or generic objects from a data base.

5. POINT CLOUDING

Technically point cloud is a database containing points in the three-dimensional

coordinate system. However, from the typical workflow perspective, the only
important thing is, that point cloud is a very accurate digital record of an object or
space. It is saved in form of a very large number of points that cover surfaces of a
sensed object.
Points in a point clouds are always located on the external surfaces of visible objects,
because this are the spots, where ray of light from the scanner reflected from an
objects.
If individual points size is big enough in the certain view or zoom setting the point
cloud could be perceived as the continual surface. If the distance between points is
slightly larger, then it’s size, then we can clearly see, that this image is made of
individual points, but still, our brain can relatively easily pick up shape s of an object.
It is essential to understand that the point cloud is a set of individual, unrelated points
with defined position and colour. This makes point clouds quite easy to edit, display
and filter.
Using individual, unrelated points is a key to point clouds usefulness, because points
are objects that are easiest to handle a large amount of. A computer does not have to
care about scale, rotation and relations to other objects. Only position and colour are
things that matter for computation. This makes point clouds quite easy to edit, display
and to filter data The only real downside in the point cloud interpretation process is
that in most cases requires human involvement. Some pieces of software are certainly
capable of finding certain patterns and features but still, they cannot easily categorize,
and covert contents of a point cloud.

9
 5.1. ACQUISITION OF POINT CLOUDS
The key factor in acquiring point cloud data is the access/visibility to scanned
surfaces. It is important to remember, that point cloud is obtained by visible access
to real objects. Regardless of the method of acquisition (scanner or photos). It is
impossible to obtain points on the surfaces that are not visible from the position
from which we collect data. This means that to cover all objects that have to
combine many scanning positions. Using term density to describe resolution on
the collected dataset this usually means the distance from a point to point. Less
dense point clouds are obviously much quicker to capture.
Most point cloud databases contain not only position of a point, but also a
description of visual properties, as a colour of an object or it’s reflectivity. All of
such may or may not be included within point cloud and they are yet another factor
that affects the time the time of acquiring scan.

6. MESHING

6.1. USING TRIANGULATED MESH CREATED FROM SCANNING

POINT CLOUDS
The first technique is to use a triangulated mesh of a fractureIf the orientation of
each triangle is plotted on a stereonet, then the scatter about the mean orientation
of the fracture gives information on the dilatation angle. In the classic saw-toothed
fracture analyzed by Patton, the dilatation angle is defined as the rise angle of the
saw teeth compared with the mean orientation, as shown in . The dilatation angle
is directly related to the additional friction angle due to roughness, and on a
stereonet, the dilatation angle can be directly determined by the angle between the
mesh triangle orientation and the mean orientation of the fracture. The example in
shows a scatter of triangle orientations, with the mean fracture orientation at the
centre of the scatter.
The stereonet in is marked off in degree increments of 10 degrees, and indicates
dilatation angles ranging from a few degrees to over 30 degrees. Also the shape
of the scatter in the stereonet is elliptical, indicating roughness anisotropy
(dilatation angle varies with direction). By varying the triangle size of the mesh,
scale-dependent roughness can be determined. As an important note, the triangle

10
 size needs to be greater than the scanner error, or else roughness due to
measurement error will be calculated.

7. SITE MAPPING

To evaluate the potential application of laser scanning, it is important to establish

what is the typical procedure for site mapping in an engineering project, and then find
out if laser scanning can help to improve or solve the problems of the traditional
methods.
For a site mapping, a typical procedure is:
1. Collect the raw data in the field
The raw data consist of different types, which depends upon the different projects, but
all projects must collect some data, such as location, geometry (of fracture, rock
surface etc.), through measurement, photographs, taking notes of the observations,
etc. Digital methods can make this step much faster and accurate than the traditional
approaches. More detailed description in relation to this subject will be presented in
the later sections.
2. Calculate and analyse the raw data
After the field data collection, the raw data must be processed for different purposes,
such as calculation of fracture orientation and plotting on a stereogram for fracture
analysis and stability analysis, importing the fracture parameters for numerical
modelling, and also for the design of a dam or tunnel. All of the data processing is
now computer-aided, so the raw data must be digitized, which is timeconsuming with
the traditional recording method.
3. Present and transform the processed data
The processed data and results, either in table or in graphical form, should be
presented in a way, or exported to other software, to enable further analysis, modelling
or design, which might be in digital form and computer-aided.

4. Store and archive the raw data and results

With the development of computer science, the data and results are now mostly stored
and archived in a digital way on CD/DVD, band or external/internal hard disk, so it
is important to convert all the data to digital form. So, in order to avoid the drawbacks
of the traditional methods, a new method must have the following important features:

11
  Quickly capturing the data in the field.
 Digitally collecting the data in order to utilise the computer resources to speed
up the procedure of data capturing and processing.
 Having the ability to visually operate the data so that the operator’s
background knowledge and experiences can be fully utilised to observe the
complicated phenomena related to a tunnel, and then obtain the required
information for engineering applications.
 Keeping a certain level of accuracy for different engineering applications.
 Possibility to capture the data in 3d without physically contacting rock faces
over a range of distances.
As 3D laser scanning has many advantages for 3D measurement and documentation,
it shows great potential for application to different projects in engineering.

8. TLS FOR TUNNELS

8.1. FACTORS CONSIDERED

Conditions underground are not the most suitable for conducting measurements:
it might be dark, damp, and dusty. We can't do measurements during active
construction work but on the other hand, the work cannot be stopped for very long
time for survey. Yet while constructing new underground spaces or maintaining
the old ones, there are many applications that require accurate information of the
precise dimensions of the tunnel:
 The thickness of the cement surface layer sprayed on the rock face must be
monitored
 Structural discontinuities need to be evaluated and documented
 Rock mass and the surface analysed
 Deformations and seepage must be kept eye on
 As well as construction quality needs to be monitored

8.2. TUNNEL SCANNING

Terrestrial static or mobile laser scanning offers the fastest, safest and most
accurate way of gathering data for these tasks. LiDAR requires no external light

12
 and the equipment is robust enough to cope the underground conditions. Static
scanners are portable and mobile scanner can be mounted on a car or train or other
vehicle. They produce a dense 3D point cloud with high speed. Modern
positioning systems provide accurate and reliable positioning even during quite
long GPS outages. Surveying railway or motorway tunnels can be conducted with
mobile scanner without disrupting the traffic flow or placing personnel at risk in
the middle of traffic.
Laser scanning produces a dense 3D point cloud depicting the surface of the tunnel
walls as well as all objects inside the tunnel. Most scanners also provide intensity
information for each point, thus giving information of the geological structure and
the surface material of the wall. Additionally, in TerraMatch the intensity
information can be utilized in combination with bright control signals for mobile
scan line matching purposes. More data can be collected by taking accompanying
visual-light-based camera images at the same time.

9. MEASUREMENT AND ANALYSIS

Although there are lots of applications using the terrestrial laser scanning the
measurement planning is seldom considered and mostly limited to cost minimization
by scanning to the maximum range of the scanning equipment. However, other factors
besides distances that substantially influence the accuracy of the measurements tend
not to be taken into account when planning data capture such as the size of the laser
footprint the incidence angle, the scanning density, and the geometry of the tunnel
itself．
Argüelles-Fraga studied the impact of the location of the scanner, the incidence angle,
the footprint size and the scanning density on geometric verifications of circular cross-
section tunnels (Argüelles-Fraga et al). He also suggested a data capture methodology
designed to optimize tunnel scanning tasks using TLS. His methodology is based on
theoretically estimating the position of the laser spot on the tunnel surface. The
distance from the scanning position to each point on the tunnel wall can be
theoretically calculated with the radius of the circular section, the laser scanner height,
the distance from the tunnel centre to the scanner, and the laser scanner measures
horizontal and vertical angle

13
Argüelles-Fraga also discussed other three factors affecting measurement accuracy:
point density, incidence angle, and footprint size． Considering with these four
factors, he did an approach to determine the distance and angular intervals which yield
a specific accuracy for a minimum scan time． In order to validate his method, a
simulated tunnel scan and real tunnel scan were carried out. The results visually
showed the influence of the factors above on the scan accuracy．In other words,
tunnel geometry, scanner position， scanning density， incidence angle， and
footprint size are closely related factors that could affect the results of the
measurements．
Pejic gave an optimal solution for surveying tunnel geometry using the laser scanning
technology to reliably inspect railway tunnels and create as built documentation．
This methodology provides the optimization of scanning parameters, scans
registration， georeferencing approach and the survey control network design． The
maximal size of the scanner shifting along the tunnel alignment is primarily
conditioned by the factors including the incidence angle of the laser beam and the
point density distribution． Pejic introduced the so-called arbitrary georeferencing
approach in the long tunnel scanning that controlled the point cloud geometric
distortions to the required limits and contributed to time and material resources
savings． The optimal design of the survey control network ensures the required
positional accuracy and the reliability of the measurements, together with a cost
effective approach to the tunnel surveying．
This method was tested by the empirical results of the modelling and profiling of 12
tunnels in a single track railway． The lengths of these tunnels were from 60 m to
1260 m, with a total length of 3.5 km． Due to the specific geometry of the case study
tunnels the maximal favorable laser incidence angle was 78° with a distance of 13m,
and consequently the optimal size of the scanner shifting along the tunnel alignment
was 26. The survey control network was designed with the condition that the optimal
reliability factors were within the required limits for engineering networks. A priori
estimation of the control network positional uncertainty and posteriori adjustment
results show that the achieved positional accuracy of the control points is
approximately five times higher than the requested absolute accuracy of the tunnel
model. On the largest tunnel example, it was shown that the arbitrary georeferencing

14
 approach assured that the optimal registration error size was within the requested
limit.

Based on Argüelles-Fraga's study, Javier proposed the influence analysis of range and
angle of terrestrial laser scanning measurements incidence on tunnel inspection. In
this study, a methodology to build an error model of TLS measurements was
suggested. Simulating errors on the point cloud measured with a TLS system is
possible to analyze the effect of the errors due to the distance to the object and the
angle of incidence on tunnel inspection． For the maximum distance recommended
for tunnel inspection, the errors are mainly due to the angle of incidence．
Conclusions concerning the influence of the position of the scanner on errors were
also extracted. Mostly, the suitable position should be the center of the tunnel, the
error distribution along cross-sections is homogeneous, and the maximum errors are
fewer than that when the scanner near the tunnel gable (a common position in practical
situations). When the scanner is near the tunnel gable, the error distribution is not
homogeneous along the same cross-section and greater errors would occur on the wall
near the scanner, due to the angle of incidence. However, when a surface is fitted to
the point cloud, these inconveniences are countered in tunnel inspection due to the
higher density of points on the wall near the scanner．

10. ADVANTAGES AND DISADVANTAGES

10.1. ADVANTAGES
The advantages are:
 Reduced project costs.
 Avoids missed schedules.
 Less operation and maintenance inefficiencies and problems.
 Low safety problems.

10.2. DISADVANTAGES
 Laser 3D data sets requires post processing to produce a useable
output.
 No common data exchange format currently in use.

15
  Difficult to stay up to date with developments.
 Output requires manipulation to achieve acceptable recording quality

11.OTHER APPLICATIONS

 Deformation monitoring
 Distance measurement
 Determine building maintenance requirements
 Determination of earthwork quantities

12.CASE STUDY

12.1. LASER SCANNING OF UNDERGROUND UTILITY TUNNELS

Location: Ann Arbor, Michigan
Task: To map out the existing location and condition of underground utility
tunnels in order to chart the locations of the utilities and to document the tunnels'
elevations, slopes, and structural features. Challenge: Many of these tunnels were
built in 2009 were not well-documented; therefore, it is difficult to know the
tunnels' paths and what is in each section. Due to the conditions inside the tunnel
system, including tight spaces, hard-to-reach areas, hard-to-see pipes, and the
sheer expanse of the tunnel system, it was not feasible to manually map the tunnel
structure and its contents. Solution: TruePoint's use of 3D laser scanning made it
possible to capture information in the hardto-reach and hard-to-see places.
TruePoint then took the point cloud data from the scans and turned it into a layered
3D CAD model. This allows the user to analyze the data by turning off layers,
such as walls, that were impeding the view of pipes and other utilities located
inside the tunnel system.
This method allows for the unique ability to see parts of the tunnel that are not
ordinarily visible due to restricted view and/or area access. Additionally, by tying
the scans to survey control both inside and outside the tunnel system, the user is
now able to study the slope and elevations of the tunnel systems, as well as where
these tunnels are located on state plane coordinates.

16
13. CONCLUSION
Terrestrial laser scanning is increasingly being used for tunnels, becoming a powerful
tool for both construction and maintenance management. In near future, the laser scanning
will probably be a standard tool for geologic and geometric analysis for tunnels. As the
technique is also progressing, more accurate and inexpensive TLS devices will appear
with more extensive applications. Nevertheless, the huge amount of data will remain a
problem since the computers need to be more powerful with increasing data acquisition
capacity as it already needs hours to process the data for some simple applications． The
real challenge is to develop some advanced methods with more complex image
processing algorithms. In most of the current applications, it is necessary to process the
scan data with some simple algorithms, ignoring lots of potential information. There is
the possibility to develop a standard workflow for the optimal use of the capacities of
laser scan instruments. There is still a huge potential to excavate in tunnel data in the
future．

17
REFERENCES
[1] Wang, W.X.; Zhao, W.S.; Huang, L.X.; Vimarlund, V.; Wang, Z.W., 2017,
Applications of terrestrial laser scanning for tunnels: a review. J. Traffic Transp.
Eng. (English Edition), 1, 325–337.
[2] Gikas V, 2015, Three-dimensional laser scanning for geometry documentation
and construction management of highway tunnels during excavation. Sensors, 12,
11249–11270.
[3] Kong X L, Ou B, 2015, The application and research of 3D laser scanning
technology in finish tunnelling survey. Urban Geotechnical Investigation ＆
Surveying, (2): 100-102
[4] D-L. Chen1,2, Y-Y. Lu1*, Y-M. Chen1, L. Ma1, D-Z Jia3, L. Cheng1, M-C. Li1,
D. Hu2, and X-F. He, 2018, Automated and Efficient Extraction of Highway
Tunnel Lining Cross-sections Using Terrestrial Laser Scanning (TLS)
[5] Chris PAGE, Pascal SIRGUEY, Richard HEMI, Ghyslain FERRE, Elisabeth
SIMONETTO, Christophe CHARLET, 2017, Terrestrial Laser Scanning for the
Documentation of Heritage Tunnels

18