MODELLING POINT CLOUDS FOR PRECISE STRUCTURAL DEFORMATION 
MEASUREMENT 
S. J. Gordon , D. D. Lichti, M. P. Stewart and J. Franke 
Western Australian Centre for Geodesy, Curtin University of Technology 
GPO Box U1987, Perth, WA, 6845, Australia 
S.Gordon G curtin.edu.au 
KEY WORDS: Laser scanning, Metrology, Modelling, Monitoring, Point Cloud, TLS 
ABSTRACT: 
Terrestrial laser scanners can rapidly acquire thousands of 3D points over a structure. The individual scan points are of relatively low 
precision (£2mm — +50mm) depending on the instrument type. However, combining the dense 3D point data with judicious 
modelling strategies can produce 
a very precise surface model. A surface model has advantages for structural deformation 
monitoring where deflections are small («50mm) and the shape change can potentially vary across the entire structure. The notion of 
the research presented in this paper is to exp 
measuring structural deformation. Two experiments ha 
loit the dense 3D point data (clouds) to assess the sensitivity of laser scanners for 
ve been undertaken where, in each experiment, a beam was subjected to 
controlled loading. The first experiment involved a timber beam (5.0m x 0.2m x 0.1 m) mounted on an indoor load-testing frame. The 
beam was subject to a maximum of 40mm of vertical deflection. T 
he focus of the second experiment was a concrete beam (7.0m x 
0.5m x 0.5m) placed on an outdoor load-testing frame where it was loaded until failure. All loading was induced by a hydraulic jack 
and occurre 
d in increments permitting measurements to be made by laser scanners. A Riegl LMS-Z210 laser scanner was used for 
both experiments and a Cyra Cyrax 2500 was available for the first. The scanner measurements were validated using close-range 
photogrammetry (accuracy of 1:40,000 of object size or better). 
1. INTRODUCTION 
The difficulty of monitoring deflections is finding a spatial 
measurement technique that encompasses numerous desirable 
properties, such as reliability, accuracy, low-cost and ease of 
installation (Stanton et al., 2003). There are many methods that 
purport some of these advantages but not all. For example, 
digital photogrammetry can be relatively inexpensive and highly 
precise; as well as offering rapid, remote, three-dimensional 
data capture and images which provide a permanent visual 
record of the test. However, the necessary use of targets may be 
disadvantageous in some circumstances, especially when the 
object is hazardous to operators or inaccessible. Furthermore, 
unless convergent imaging is practiced, the depth dimension 
can be poorly observed. This can occur when the laboratory 
lacks sufficient space to satisfy an even geometric distribution 
of exposure stations. The photogrammetric process also lacks 
scale definition, requiring measurements to be acquired using 
additional instrumentation, such as a precise scale bar. 
Traditionally, contact sensors, such as dial gauges and linear- 
variable-differential transducers (LVDTs), are employed for 
structural deflection experiments because of their high precision 
spatial measurement capabilities. However, their contact nature 
precludes them from use during the final stages of destructive 
load testing and they are only capable of acquiring 
measurements in one dimension. Importantly, the number of 
monitoring sites, or data density, is limited by the number of 
contact sensors available for the experiment. This is also true 
for target availability in photogrammetric metrology, although it 
is less of a problem since photogrammetric targets are 
inexpensive and may be quickly placed on the object of interest 
and its stable surrounds. 
e rr —— 
  
  
" Corresponding author. 
Terrestrial laser scanners (TLSs) are modern geomatic data 
capture instruments that offer numerous measurement benefits 
including three-dimensional data capture, remote and non- 
contact (i.e. targetless) operation, a permanent visual record and 
dense data acquisition. TLSs are currently being used in a 
variety of projects, including heritage mapping, as-built 
documentation and topographic surveys. However, the precision 
of TLSs is not perceived adequate for industrial metrology 
applications, such as deformation monitoring. 
The advantage of TLSs is that, although individual sample 
points are low in precision (e.g. £2mm to 50mm), modelling 
of the entire point cloud may be effective for explaining the 
change of shape of a structure. A modelled surface will be a 
more precise representation of the object than the unmodelled 
observations. In light of this notion, a methodology for 
measuring structural deformation, relying on theoretical aspects 
of beam mechanics and implemented by constrained least- 
squares curve fitting, has been developed and is presented in 
Section 2. A statistical test for assessing the redundancy of 
estimated parameters is given in Section 3. The results of two 
structural deformation monitoring experiments, involving 
beams (one concrete and one timber) being loaded in a load- 
testing frame, used to test the analytical modelling strategy are 
presented in Sections 4 and 5. Both experiments Were 
controlled with convergent digital photogrammetry. A 
discussion focussing on instrument set up is given in Section 6 
and the conclusions are presented in Section 7. 
954 
In 
Su 
SU 
ar 
de 
Imi 
m 
in 
hi; 
thi 
ar 
pr 
po 
are 
wi 
Jol 
dif 
as 
the 
dis 
to | 
Thi 
twi 
Thi 
con 
inte 
The 
Cor 
of i 
beai 
  
Supp 
The 
side 
poin 
gene 
Eq. : 
Figu 
mod 
evide 
term 
Curve