Measurement 44 (2011) 527–538

Contents lists available at ScienceDirect

Measurement
journal homepage: www.elsevier.com/locate/measurement

Laboratorial test monitoring applying photogrammetric

post-processing procedures to surface displacements
D. Dias-da-Costa a,⇑, Jónatas Valença b, Eduardo N.B.S. Júlio a
a
ISISE, Faculdade de Ciências e Tecnologia Universidade de Coimbra, Rua Luís Reis Santos – Pólo II da Universidade, 3030-788 Coimbra, Portugal
b
ISISE, Department of Civil Engineering, Polytechnic Institute of Coimbra, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: Most of the traditional laboratorial methods for monitoring displacements and strains at
Received 7 July 2010 the surface of the specimen present several constraints, namely: (i) limitations in hardware
Received in revised form 4 November 2010 positioning; (ii) costly equipment and human resources; and (iii) time-consuming data
Accepted 17 November 2010
processing. Consequently, the development of new methods capable of eliminating these
Available online 23 November 2010
drawbacks is of utmost interest.
Herein, a new technique for laboratorial test monitoring is presented. By using photo-
Keywords:
grammetry and image post-processing, all the above mentioned drawbacks are overcome.
Photogrammetry
Post-processing
Furthermore: (i) both displacement and strain fields can be monitored at a practically
Strain field unlimited number of target points at any stage; (ii) it is a cost effective method, since data
Crack is acquired with non-professional digital cameras; and (iii) it is a fast procedure since data
Test monitoring is automatically processed. Additionally, high precision is reached allowing an accurate
characterisation of the fracture localisation process, including the establishment of a cor-
relation between the latter and the localisation of cracks and their evolution in time.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction More recently, new techniques have been developed.

Monitoring of most common civil engineering laborato-
rial tests is still performed with traditional methods and
devices, namely: (i) LVDTs to measure displacements;
and (ii) strain gages to measure the strain field. These de-
vices present several limitations related to: (i) positioning,
due to a reduced number of devices usually available and/
or due to lack of space in critical zones of the specimens;
(ii) high cost, normally further increased by the need of
costly hardware, such as data loggers; and (iii) time-
consuming procedures. Moreover, and as a consequence,
in most tests only a reduced number of points are
thoroughly monitored.
⇑ Corresponding author. Tel.: +351 239 797 258; fax: +351 239 797
259.
E-mail address: dcosta@dec.uc.pt (D. Dias-da-Costa).
Optical fibre sensors, for instance, are experiencing a grow-
ing interest [1]. Although allowing for high precision mea-
surements, all the above mentioned drawbacks found on
traditional devices and methods are still present. Further-
more, complex and burdensome cable connections are
needed for full-field strain monitoring.
Other techniques recently developed include: shearog-
raphy, digital image correlation, laser speckle analysis,
thermal imaging and laser scanning vibrometry [1,2]. With
these, positioning is no longer a drawback. However, these
equipments are expensive and their usage has been limited
due to the need for technical expertise.
In this paper, a new technique, using photogramme-
try and post-processing tools, herein designated by
‘visual-DSC’ – ‘Visualisation of Displacements, Strains and
Cracks’, is presented. The visual-DSC method was devel-
oped aiming at the following: (i) monitor a virtually unlim-
ited set of pre-defined points identified by small painted

0263-2241/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.measurement.2010.11.014
528 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538
Nomenclature
Ecm Young’s modulus
fcm,cube average compressive strength evaluated from
cubes
targets, thus avoiding the limitation of having a reduced
number of LVDTs at the laboratorial facilities and overcom-
ing placement restrictions in small areas of test specimens;
(ii) obtain a cost-effective technique by using a low cost dig-
ital camera and a laptop, avoiding the use of any other type
of costly hardware, such as a data logger; (iii) obtain both
displacement and strain fields at any stage in a fast and
reliable way by automatically storing and processing data.
Some authors have already dealt with displacement
monitoring using photogrammetry [3–8]. For instance
Sachtleber et al. [9] evaluated the displacement field and
the corresponding plastic strains, whereas Thomas and
Cantré [10] aimed at evaluating the strain field of cracked
soils. However, the strain field was still constrained to nar-
row areas and good results were achieved using metric
cameras [11,12]. Visual-DSC is capable of avoiding all these
drawbacks.
In order to calibrate visual-DSC, namely to adequately
characterise both displacement and strain fields and also
to correlate the latter with cracking, the new method
was used in a vast campaign of push-off tests. Herein,
one of these push-off specimens, monitored until failure
with visual-DSC, is presented. All capabilities of visual-
DSC are fully illustrated and a complete understanding of
the structural behaviour of the specimen can be obtained.
First, in Section 2, the visual-DSC method is described
regarding both set-up and data processing. In Section 3, de-
tails concerning the experimental set-up and the valida-
tion method are provided. In Section 4, results are
analysed and discussed. Finally, in Section 5, the most
relevant conclusions are presented.
(a)
Fig. 1. Photogrammetric set-up: (a) photo; and (b) scheme showing the targets of the specimen and location of profiles.
fctm average tensile strength
e1 first principal strain
e3 third principal strain
2. Photogrammetric procedure
2.1. Set-up
Photogrammetric data was obtained with a Nikon
D200 installed on a tripod, using 28 mm focal length lens.
Self-calibration based on the conventional bundle adjust-
ments method [13] was used to estimate the internal
parameters of the camera. Images were acquired at
maximum resolution, corresponding to an image size of
3872  2592 pixels. Conditions were created to insure a
homogeneous and diffused light pattern with natural light
sources.
A rigid frame was designed to be placed in front of the
specimen, containing 32 static targets, assumed as the
ground-truth reference (see Fig. 1). The specimen was
marked with a 25  28 mm2 grid of 194 high-contrast
circular targets which are represented in Fig. 1b. The
diameter of the targets, 10 mm, was defined considering
the maximum distance between the camera and the tar-
get, the dimensions of the CCD sensor of the camera, the
number of pixels in the image, the focal length, and also
taking into account the following constraints: (i) the
Least Squares Matching (LSM) algorithm must be capable
of automatic identification of the central point of each
target with sub-pixel accuracy; (ii) the diameter of the
targets should be small enough to avoid interference
with visual crack identification. This led to an average
diameter of 26 pixels per target. For direct comparison
with the LVDTs, four additional targets were installed
(see Fig. 1a).
(b)
 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538
2.2. Data processing
Each stage of image acquisition led to an independent
photogrammetric project containing at least four images,
taken from distinct point of views, with a multistation net-
work [14,15]. All pictures taken included all targets, the
latter located at an average distance of 2.0 m.
Just before loading, a reference photogrammetric pro-
ject, denoted stage 0, was built. During testing, at each
stage relevant to monitoring, the loading machine was
immobilised by keeping constant the vertical position of
the load plate and the corresponding set of images was
acquired. Fig. 2 contains a representation of all camera
positions at stage 1. An average convergence angle of only
41–46° was obtained due to the presence of a reaction wall
(see Fig. 2) which is acceptable according to [15].
Each stage was processed with commercial photogram-
metric software. Matching points between images were
automatically identified and processed to build a three-
dimensional model of the surveying area. The ground-
truth constraints were used for aligning the set of images
and to scale the model. Furthermore, scale bars composed
of high-contrast circular targets were placed in the steel
frame surrounding the specimen (see Fig. 1a).
For the subsequent analysis, which is performed on the
surface of the specimen, the bidimensional coordinates of
the corresponding target points are used. A Delaunay trian-
gulation was adopted [16]. This algorithm defines a set of
triangles corresponding to a set of points, such that no
point is contained in any triangle’s circumscribed circle.
Therefore, an auxiliary mesh was built establishing the cor-
responding connectivities between target points. At the
external boundary of the specimen, distorted elements
were removed. During the progressive stages of analysis,
specially when approaching failure, few target points dis-
appeared. In this case, the missing photogrammetric data
was obtained by nonlinear interpolation [17]. To prevent
Fig. 2. Photogrammetric set-up: scheme identifying the different parts.
529
this procedure from leading to a decreased accuracy at
the missing targets, a cloud containing several neighbour-
hood targets was considered. With the displacement field
defined at every target point and from the strain–nodal dis-
placement matrix of the auxiliary mesh, the strain field was
directly computed. As shown in Section 4, finite element
post-processing software was used to gather the most rel-
evant information considering structural monitoring.
3. Experimental work
One of the main research fields of the authors is the
experimental and numerical characterisation of the behav-
iour of structural interfaces between new-to-old concrete
layers [18–20]. To experimentally assess the shear strength
between the original concrete layer and the new concrete
layer with added reinforcement crossing the interface,
push-off tests are usually adopted. Herein, some of these
specimens are used, to test the visual-DSC method de-
scribed in Section 2. A detailed analysis is presented in
the following sections.
3.1. Push-off tests
The geometry adopted for the push-off specimens is the
one defined by Hofbeck et al. [21] and is represented in
Fig. 3. The push-off specimens are anti-symmetrical and
composed by two identical ‘L’ shape halves with a
254  546  127 mm3 envelope prism. Each half is rein-
forced with six S500 steel bars with 10 mm diameter and
eight S500 steel stirrups with 8 mm diameter. The rein-
forcement crossing the interface is composed of three lev-
els containing two S500 steel bars with 12 mm diameter.
The specimens were produced according to the follow-
ing procedure (see Fig. 4): (a) the reinforcing bars and stir-
rups of the substrate half are assembled; (b) the substrate
half is cast; (c) the reinforcing bars and stirrups of the
added half are assembled and two polystyrene plates on
the original half are placed to produce a 19 mm gap; (d)
the roughness of the interface surface is increased; and
(e) the added half is cast.
Fig. 3. Push-off specimen (dimensions in mm).
530 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538

Fig. 4. Different steps of production of push-off specimens: (1) assembling substrate rebars; (2) substrate casting; (3) assembling added half rebars; (4)
added half casting.

(a) (b) (c)

(d) (e) (f)

Fig. 5. Experimental test at: (a) stage 1; (b) stage 2; (c) stage 3; (d) stage 4; (e) stage 5; and (f) stage 6.
 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538 531

Different concrete mixes were adopted for the substrate

and for the added layer having average 28 day compressive
strengths, fcm,cube, of 27.2 N/mm2 and 69.1 N/mm2, respec-
tively. According to the Eurocode-2 [22], this corresponds
to an average tensile strength fcm = 1.9 N/mm2 and an
Young’s modulus of Ecm = 29000 N/mm2 for the substrate,
and fcm = 4.4 N/mm2 and Ecm = 39000 N/mm2 for the added
layer.
The push-off test was performed on a 5000 kN universal
testing machine with displacement control (see Fig. 1a). A
load cell, TML CLC-50A, placed at the top of the specimen
and two displacement transducers (LVDTs), TML SDP
200D, connected to a data logger, TML TDS 602, were used
to measure, respectively, the load applied and the defor- Fig. 7. Displacements obtained with visual-DSC vs. displacements mea-
mation of the specimen (see Fig. 1a). sured with LVDTs.

The specimen was loaded by imposing a displacement

at the lower plate of the machine at a rate of 0.015 mm/ obtained. This means that visual-DSC is suitable for identi-
s. The opposite side was hinged and therefore small rota- fying strain localisation related to crack propagation, but
tions were allowed. In Fig. 5, pictures from the same sta- the elastic strain field cannot be accurately evaluated with
tion for all stages of loading are shown. The presence of this resolution.
shaded areas is explained in Section 4.3.
4. Results and analysis
3.2. Validation
4.1. Load vs. displacement curve
According to the process described in Section 2.2, each
stage of data acquisition requires the loading machine to The load vs. average displacement curve is represented
be stopped at a constant position. This is strictly related in Fig. 8. Until stage 2, the structural response is almost lin-
to the multistation network requiring the specimen to re- ear elastic, although a crack can be identified at the upper
main static during a complete set of pictures. The average ‘arm’ (Fig. 5a). Between stage 2 and stage 3, there is an
displacement and load vs. time curves are represented in
Fig. 6. The time interval during which the photos have been
acquired is also represented in red for each stage.
The two redundant target points installed in each LVDT
were used for direct comparison with the values obtained
with the visual-DSC method (Fig. 7). This figure shows a
high correlation between both methods.
All targets were in the same photogrammetric survey-
ing conditions. Therefore, the differences between the
coordinates of the 32 static target points in each project
were assumed to be within the precision of the photo-
grammetric survey. The RMS of these differences is
0.05 mm, whereas the average value is 0.04 mm.
The precision in terms of strains was estimated assum-
ing a locally uniform strain field. Using a grid with a spac-
Fig. 8. Load vs. average displacement curve.
ing of 25 mm, an average strain precision of 0.2% was

(a) (b)
Fig. 6. Representation of: (a) displacement; and (b) load vs. time curves.
532 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538

(a) (b) (c)

(d) (e) (f)

Fig. 9. Displacement field of the target points at: (a) stage 1; (b) stage 2; (c) stage 3; (d) stage 4; (e) stage 5; and (f) stage 6.

important loss of stiffness and a sudden load drop when a
new discontinuity appears at the upper ‘arm’ of the speci-
men – see Fig. 5b and c. This crack further develops being
easily identified at stage 4 – see Fig. 5d. From stage 3 to 4,
two significant drops are again related to the development
of cracks. After stage 4, no sudden changes on the applied
load occur. This is due to the fact that the localisation pro-
cess ends and the gradual opening of the existing active
cracks are observed – see Fig. 5d and e. Finally, the speci-
men fails by bending of the upper ‘arm’ since the interface
is heavily reinforced (see Fig. 3).
4.2. Displacement field
the location of each target, representing the magnitude and
direction of the displacement undergone since stage 0.
Table 1 contains the corresponding maximum and mini-
mum values of the absolute displacement.
From stage 1 to 5, an almost constant vertical dis-
placement for all target points located in the lower
‘arm’ of the specimen is observed. Relatively to the upper
‘arm’, a progressive counter clockwise rotation allowed
by the hinged plate of the loading machine is verified,
specially from stage 3 to 5. From stage 3 to 4 a change
occurs in the displacement field. The change is related
with the new crack opening, at the upper ‘arm’ of the
specimen. This is directly noticed by the difference

In Fig. 9 the displacement field is represented at all tar-

get points for all stages. In this figure arrows are placed at Table 2
Adopted scale for the first principal strain maps and
Table 1 corresponding maximum values.
Maximum and minimum absolute displacement for
each stage. Scale Stage Max. e1
3
Stage # Displacement (mm) 1 7.70  10
2
Min. Max. 2 1.31  10

1 2.6 5.2 2
3 2.94  10
2 2.7 6.4
1
3 3.1 8.4 4 1.01  10
4 2.2 17.2 1
5 1.75  10
5 2.1 23.2
1
6 0.1 17.4 6 1.48  10
 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538 533

#1 #3

#2

(a) (b) (c)

Fig. 10. Stage 1: (a) first principal strain; (b) map of cracks; and (c) details of the specimen.

between vertical components of the displacement vectors 4.3. Strain field

occurring at this area – compare also Fig. 5c and d. Fur-
thermore, the central part of the specimen undergoes a
global rotation to accommodate the opening of those dis-
continuities. This behaviour is further emphasised in
stage 5, where the upper right part of the specimen is
gradually separated from the remaining body – compare
Figs. 5e and 9e. Finally, the unloading of the specimen
occurs at stage 6, with an almost complete recovery of
the displacement at the lower part of the specimen.
However, the remaining part reveals important
permanent displacements. A small sliding at the interface
can also be pointed out.
Table 2 contains the intervals adopted in Figs. 10–14
and in Fig. 16 for the first principal strain maps and
corresponding maximum values. In Fig. 15 the third princi-
pal strain map is shown only for stage 5. All strain field
maps contain superimposed crosses, identifying the first
and the third principal directions represented in red and
black, respectively. Furthermore, in order to assist a com-
plete understanding of the structural response, the crack
pattern obtained by hand drawing and some details of
the specimen are also shown for each stage. These details
correspond to the shaded areas in Fig. 5.

#1 #3

#2

(a) (b) (c)

Fig. 11. Stage 2: (a) first principal strain; (b) map of cracks; and (c) details of the specimen.
534 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538

#1 #3

#2

(a) (b) (c)

Fig. 12. Stage 3: (a) first principal strain; (b) map of cracks; and (c) details of the specimen.

According to Section 3.2, the strain precision of the
method is 2  10 3, which is enough for identifying the
strain localisation process, as it is shown next.
During stage 1, there are several localisation areas rep-
resented in Fig. 10a. These are in agreement with the crack
pattern from Fig. 10b. Details #2 and #3 emphasise the fact
that these are still incipient cracks. However, at the upper
right of the specimen, and according to the strain map, an
important crack is already formed – see Fig. 10a and detail
#1. Furthermore, crack orientation is well described by the
first principal strain direction, represented in red, for all
localisation areas – compare Fig. 10a and b.
According to the first principal direction, the upper
right horizontal crack extends and curls towards the exte-
rior at stage 2 – see the agreement between Fig. 11a and b
and detail #1. The vertical crack at the interface also devel-
ops at this stage – detail #3 and Fig. 11a. There is a growth
of the area of strain localisation identified at the top of the
specimen which is related to the formation of a vertical
crack (see Fig. 11b). Furthermore, a small discontinuity
forms at the bottom left of the specimen (see detail #2).
At stage 3, almost all previously identified discontinu-
ities and localisation areas grow: (i) the crack located at
the interface now extend along the specimen – Fig. 12a
and b, detail #3; (ii) the horizontal crack at the bottom left
of the specimen develops; and (iii) the crack of the middle
top of the specimen opens, becoming more visible – see
Fig. 12a and detail #2. It is observed a new crack opening

#1 #3

#2

(a) (b) (c)

Fig. 13. Stage 4: (a) first principal strain; (b) map of cracks; and (c) details of the specimen.
 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538 535

#1

#2

(a) (b) (c)

Fig. 14. Stage 5: (a) first principal strain; (b) map of cracks; and (c) details of the specimen.

at the top right of the specimen, parallel to the existing
one, seen in detail #1. This leads to an increased strain rep-
resented by darker areas in Fig. 12a.
At stages 4 and 5 the opening of most of all previously
formed discontinuities occurs, see Figs. 13 and 14. The
interface of the specimen does not develop (compare
Figs. 13a and 14a), thus remaining practically unchanged
during these stages. This is also in agreement with the
crack patterns represented in Figs. 13b and 14b. It is
stressed out that the darker areas of the strain map spread
significantly in the left bottom and top right, forming the
failure mechanism consisting: (i) on the development of
an hinge and consequently opening of the horizontal dis-
continuities in the upper right area induced by rotation;
and (ii) on the crushing of the bottom left part of the spec-
imen, where the load concentrates – see Fig. 15a and detail
#4. Also, because of the rotation of the upper part, the
upper left gap of the specimen closes crushing the edges
– Fig. 15a and detail #3.
Finally, at stage 6 unloading of the specimen occurs and
the residual deformation is represented in Fig. 16a. It can
be concluded that, although in Fig. 9f the lower part of
the specimen does not reveal significant displacements at
the bottom part, important permanent damage is there
present.
From this section it can be concluded that the disconti-
nuities open perpendicularly to the first principal tensile
strain, according to the Rankine criterion. This is stated
by the agreement found between the crack map and the
first principal tensile strain direction. Furthermore, there

#3

#4

(a) (b) (c)

Fig. 15. Stage 5: (a) third principal strain; (b) map of cracks; and (c) details of the specimen.
536 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538

#1

(a) (b) (c)

Fig. 16. Stage 6: (a) first principal strain; (b) map of cracks; and (c) detail of the specimen.

(a) (b)
Fig. 17. First principal strain measured along: (a) profile 1; and (b) profile 2.

is a close relation between cracking of the specimen until  in stages 4 and 5, the active discontinuity develops sig-
failure and the strain localisation areas identified by means nificantly (see detail #1 from Figs. 13c and 14c);
of the strain maps.  in stage 6 only a small recovery is verified by unloading.

4.4. Crack opening Profile 2, which is depicted in Fig. 17b, shows a mono-
tonic increase of the strain in progressive stages. During
The analysis presented in the previous section can be unloading almost no strain recovery is observed.
extended by computing strain profiles. For instance,
Fig. 17 contains the first principal strain evaluated along
the profiles defined in Fig. 1b.
Fig. 17a is related to the profile 1 and gives the follow-
ing information:

 is possible to identify the strain since stage 1 as well as

its further development in stage 2. At both stages only
one crack is developing in accordance to detail #1 from
Figs. 10c and 11c;
 at stage 3 a new discontinuity is formed which becomes
the active discontinuity (see detail #1 from Fig. 12c).
This is seen by the relative jump in the strain in rela-
tion to the previous crack that, at stage 4, slightly
closes; Fig. 18. Crack opening along each profile for all stages.
 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538
(a)
Fig. 19. (a) Crack magnifier vs. image processing measurement of the crack width rulers; and (b) crack opening obtained with visual-DSC vs. image
processing, along profile 1, for all stages.
Finally, by integrating the strain along the profile it is
possible to evaluate the opening of all intercepted discon-
tinuities. Fig. 18 presents that evaluation superposed with
the error. This can be roughly estimated by multiplying
the strain precision by the length of the cut. The disconti-
nuity opening for profile 2 is not represented for both
stages 1 and 2 since the error is greater than the measured
value.
Verifications were performed by image processing
using the crack width rulers that can be seen on both sides
of the tested specimen in Fig. 1a, composed by a set of
lines with increasing thickness. It is stressed out that this
parameter has first been measured using both a crack
magnifier and image processing to validate the latter. In
Fig. 19a a linear correlation between these two techniques
is presented, showing a correlation factor of approxi-
mately 1.0, thus certifying the use of image processing
to convert pixels into millimetres. Next, the opening of
all cracks crossed by profile 1 was computed by using both
image processing and visual-DSC to validate the latter. In
Fig. 19b, a linear correlation between these two tech-
niques is presented, showing a good match between both
procedures, with a correlation factor of 0.993, this way
also certifying the use of visual-DSC to measure crack
openings.
5. Conclusions
It has been shown that visual-DSC, based on photogram-
metry and image post-processing of data acquired with
non-professional digital cameras, is a valuable laboratorial
technique. In fact, visual-DSC is capable of characterising
both displacement and strain fields in a practically unlim-
ited set of pre-defined points, without LVDTs placement
restrictions, with a reduced equipment cost, in a fast and
reliable way. Furthermore, since there is no restriction con-
cerning large strains, the process of crack formation, the
evaluation of crack orientation as well as the correspond-
ing opening can also be obtained. It is stressed out that a
complete understanding of the structural response of the
structure until failure is achieved, totally unviable with
traditional methods.
537
(b)
Acknowledgments
J. Valença acknowledges the financial support of both
the Portuguese Science and Technology Foundation (FCT)
and EC+A – Projectos Lda, Ph.D. Grant Number SFRH/
BDE/15660/2007.
References
[1] Y.H. Huang, L. Liu, F.C. Sham, Y.S. Chan, S.P. Ng, Optical strain gauge
vs. traditional strain gauges for concrete elasticity modulus
determination, Optik – International Journal for Light and Electron
Optics 121 (2009) 1635–1641.
[2] F. Marazzi, P. Tagliabue, F.M. Corbani, Traditional vs. innovative
structural health monitoring of monumental structures: a case
study, Structural Control and Health Monitoring, (2010).
[3] C. Fraser, B. Riedel, Monitoring the thermal deformation of steel
beams via vision metrology, ISPRS Journal of Photogrammetry &
Remote Sensing 55 (2000) 268–276.
[4] T. Whiteman, D. Lichti, I. Chandler, Measurement of Deflections in
Concrete Beams by Close-Range Photogrammetry, in: Symposium on
Geospatial Theory, Processing and Applications, Ottawa, 2002.
[5] D. Jáuregui, K. White, C. Woodward, K. Leitch, Noncontact photo-
grammetric measurement of vertical bridge deflection, Journal of
Bridge Engineering (2003) 212–222.
[6] H.-G. Maas, U. Hampel, Photogrammetric technique in civil
engineering material testing and structure monitoring, Photo-
grammetric Engineering & Remote Sensing 72 (2006) 1–7.
[7] R. Jiang, D. Jáuregui, K. White, Close-range photogrammetry
applications in bridge measurement: literature review,
Measurement 41 (2008) 823–834.
[8] A.M. Ferrero, M. Migliazza, R. Roncella, G. Tebaldi, Analysis of the
failure mechanisms of a weak rock through photogrammetrical
measurements by 2D and 3D visions, Engineering Fracture
Mechanics 75 (2008) 652–663.
[9] M. Sachtleber, Z. Zhao, D. Raabe, Experimental investigation of
plastic grain interaction, Materials Science and Engineering 336
(2002) 81–87.
[10] H. Thomas, S. Cantré, Applications of low-budget photogrammetry
in the geotechnical laboratory, The Photogrammetric Record 24
(2009) 332–350.
[11] O. Koelbl, Metric or non-metric cameras, Photogrammetric
Engineering and Remote Sensing 42 (1976) 103–113.
[12] A. Meneguette, Evaluation of metric camera photography for
mapping and co-ordinate determination, The Photogrammetric
Record 11 (1985) 699–709.
[13] S.I. Granshaw, Bundle adjustment methods in engineering
photogrammetry, The Photogrammetric Record 10 (1980) 181–207.
[14] J. Mills, D. Barber, Geomatics technique for structural surveying,
Structural Surveying Journal of Surveying Engineering 130 (2004)
56–64.
[15] A. Mokarrami, H. Ebadi, Evaluation of Geometrical Accuracy
in Displacement Monitoring of Engineering Structures Using
538 D. Dias-da-Costa et al. / Measurement 44 (2011) 527–538
Close-Range Photogrammetry, in: XXI Congress The International
Society for Photogrammetry and Remote Sensing, Beijing, China,
2008.
[16] C.B. Barber, D.P. Dobkin, H. Huhdanpaa, The quickhull algorithm for
convex hulls, ACM Transactions on Mathematical Software 22
(1996) 469–483.
[17] O.C. Zienkiewicz, J.Z. Zhu, The superconvergent patch recovery and a
posteriori error estimates: part 1: the recovery technique,
International Journal for Numerical Methods in Engineering 33
(1992) 1331–1364.
[18] E.N.B.S. Júlio, F.A.B. Branco, V.D. Silva, J.F. Lourenço, Influence of
added concrete compressive strength on adhesion to an existing
concrete substrate, Building and Environment 41 (2006) 1934–1939.
[19] P.M.D. Santos, E. Júlio, Development of a laser roughness analyser to
predict in situ the bond strength of concrete-to-concrete interfaces,
Magazine of Concrete Research 60 (2008) 329–337.
[20] D. Dias-da-Costa, J. Alfaiate, L.J. Sluys, E.N.B.S. Júlio, Towards a
generalization of a discrete strong discontinuity approach, Computer
Methods in Applied Mechanics and Engineering 198 (2009) 3670–
3681.
[21] J.A. Hofbeck, I.O. Ibrahim, A.H. Mattock, Shear transfer in reinforced
concrete, ACI Structural Journal 66 (1969) 119–128.
[22] CEN, EN 1992-1-1: Eurocode 2: Design of Concrete Structures – Part
1-1: General Rules and Rules for Buildings, in: European Committee
for Standardization (CEN), 2004.