Professional Documents
Culture Documents
Xia 2012
Xia 2012
Xia 2012
DOI 10.1007/s11340-012-9687-0
Received: 22 January 2012 / Accepted: 5 October 2012 / Published online: 23 October 2012
# Society for Experimental Mechanics 2012
DIC measurements have been made using a single camera perspective difference between the images and uncertainties
[12–15]. Using a single camera is beneficial because of the associated with calibrating the angles in the geometrical
space limitations and cost reduction in setting up the system. arrangement [20], resulting in a simpler and more compact
Furthermore, for dynamic measurements synchronization arrangement. This method can be adapted and amenable for
issues of the two cameras need to be overcome. In Tong implementation with an optical microscope and thus en-
[12], a 3D surface geometry and strain measurement setup abling 3D displacement measurements of materials and
using a single camera and a rotation/translation stage was structures at the micro and nano scales (e.g., nano compo-
described. The measurement principle of the setup was sites, MEMS). Other potential applications include local
based on parallel or pinhole perspective projections of 3D strain measurements in deforming materials at the grain
objects at different orientations. Tay et al. [13] and Quan et scale under quasi-static and dynamic loading conditions.
al. [14] developed a method that uses global two- The remainder of the paper is organized as follows. The
dimensional DIC with a single camera to measure large theory for the proposed 3D-DIC experimental method is
three-dimensional displacements. They demonstrated the presented in Section 2. In Section 3, the details of the
accuracy of the proposed method on a uniformly displaced experimental setup are described. The results demonstrating
silicon wafer and a non-uniformly deformed cantilever the new 3D-DIC technique are presented in Section 4. The
beam. Pankow et al. [15] developed a 3D-DIC method discussions on the applicability and limitations of the tech-
capable of measuring dynamic out-of-plane displacements nique are given in Section 5 and the conclusions for the
using a series of mirrors and a single high-speed camera. study are drawn in Section 6.
The series of mirrors convert one single camera into two
virtual cameras that view the specimen surface from two
different angles and capture two images simultaneously. The Theory
method has the ability to obtain data at high speeds with
higher resolution than available on the market, without the The newly developed DAIC experimental setup involves
limitations related to synchronization of the two cameras. viewing the region of interest under monochromatic illumi-
Furthermore, destructive tests can be performed without the nation through a transmission diffraction line grating, as
risk of destroying the cameras. The method has been applied shown in Fig. 1. When an initially flat object (AB) is placed
successfully in static as well as dynamic shock tube experi- parallel to the grating and observed in the direction perpen-
ments on sandwich panels. A general limitation of the dicular to the grating surface, multiple views of the object
current 3D-DIC technique is the acquisition of images at are obtained corresponding to different diffraction orders.
different angles results in shape distortion of the regions that
is being interrogated. It is also necessary to employ a 3D-
d±
DIC algorithm that is needed to compute the displacements
from the two images obtained from viewing the specimen at
d
two different angles. -
A
In the present work, a new 3D-DIC technique, namely, -
Diffraction Assisted Image Correlation (DAIC), has been θ P
developed. The DAIC method utilizes a transmission dif- B
-
fraction grating and a single camera, and is capable of 3D
displacement measurement with a 2D-DIC algorithm. Dif- A
e//
fraction of light, which occurs when a light wave is incident en P
and bends around an object, has found wide use in experi-
mental mechanics for deformation characterizations e⊥ B
[16–19]. In this study, we exploit the light splitting and
bending characteristics of a diffraction grating to obtain all A+
three displacement components of a specimen from the in- θ
P+
plane displacement components of higher-order diffraction
images. The distortions in the images are minimal since the B+
specimen itself is not viewed at an angle. The validity and Transmission
accuracy of the method have been demonstrated through grating
measurement of displacements of a small thin membrane.
Fig. 1 Schematic of the optical arrangement for 3D displacement mea-
The use of the grating in place of stereoscopic images by surement by DAIC using a diffraction grating and a single camera. e⊥ and
viewing the region of interest at two different angles [8, 9] e// are unit vectors perpendicular and parallel to the grating rulings,
or using mirrors [15] to create multiple images eliminates respectively, and en is a unit vector normal to the grating surface
Exp Mech (2013) 53:755–765 757
The focus here is on the positive (+1) and negative (−1) where ðuP en Þ is the out-of-plane displacement of the
first-order images, which correspond to the first-order dif- point P.
fraction of the transmitted light. The two first-order images Equation (6a,b) imply that, if the in-plane displacement
can be thought of as virtual images formed from two virtual fields of the two virtual images are provided, the 3D dis-
objects (A+B+ and A−B−). By using a backward ray-tracing placement field of the real object can be calculated by add-
method (see Appendix), the distance between the grating ing or subtracting equation (6a) and (6b) as follows:
and the two virtual objects, d±, is found to be
uInplane
þuInplane
d uInplane ¼ P Pþ
;
d ¼ ; ð1Þ
P 2 ð7a; bÞ
cos3 θ ðuP uPþ Þe?
Inplane Inplane
uP en ¼ 2 tan θ :
in which d is the distance between the real object and the
The above equation (7a,b) form the basis of performing
grating, and θ is the first-order diffraction angle. From the
3D displacement measurement with a single camera and a
diffraction grating equation, θ is obtained as
2D-DIC algorithm. Note that adding the net in-plane dis-
1 l
placements from the positive (+) and negative (−) first-order
θ ¼ sin ; ð2Þ diffracted images removes the contribution from the out-of-
p
plane displacement, while subtracting them provides only
where λ is the wavelength of the light source and p is the the contribution due to the out-of-plane displacement. Now
pitch of the grating (1/Line Density). consider a Cartesian xyz-coordinate system with its principal
Consider a point P on the real object with a position axes (ex, ey, ez) coincident with (e⊥, e//, en). When the ruling
vector XP. The position vectors of the corresponding points on the grating is aligned in the y-direction, the three dis-
(P+ and P−) mapped onto the two virtual objects are placement components (u,v,w) of the object along the x-, y-
and z- directions are derived from equation (7a,b) as
XP ¼ XP ðd d Þen d tan θ e?
ð3Þ
¼ XP d cos13 θ 1 en d tan θ e? ðU 1 ðx;yÞþU þ1 ðx;yÞÞ
uðx; yÞ ¼ 2 ;
ðV 1 ðx;yÞþV þ1 ðx;yÞÞ ð8a; b; cÞ
where e⊥ and e// are unit vectors (in-plane) perpendicular vðx; yÞ ¼ 2 ;
and parallel to the grating rulings, respectively, and en is a ðU 1 ðx;yÞU þ1 ðx;yÞÞ
wðx; yÞ ¼ 2 tan θ ;
unit vector (out-of-plane) normal to the grating surface.
Now a displacement of P, uP, is introduced. The current
in which ðU 1 ðx; yÞ; V 1 ðx; yÞÞ and ðU þ1 ðx; yÞ; V þ1 ðx; yÞÞ
position of P becomes xp ¼ Xp þ up , and the distance
denote the x- and y-component displacement fields of the
between P and the grating becomes d up en . Then
negative and positive first-order images, respectively.
the current positions of P+ and P− are calculated as
1 Experimental Procedure
xP ¼ xP ðd uP en Þ 1 en ðd uP en Þ tan θ e? :
cos3 θ
ð4Þ Experimental Setup
Subtracting equation (3) from equation (4) gives the
The schematic of the experimental setup is shown in Fig. 2.
displacements of P+ and P− as
A white light source filtered with a narrow band optical filter
(632.8±1.0 nm bandpass, NT65-753, Edmund Optics, Bar-
1
uP ¼ uP þ ðuP en Þ 1 en ðuP en Þ tan θ e? : rington, NJ) was used to provide quasi-monochromatic illu-
cos3 θ
mination. A transmission grating of 110 lines/mm groove
ð5Þ density (NT46-073, Edmund Optics, Barrington, NJ) was
The components of uP and uP in the grating plane used to achieve multiple views of the area of interest on the
can be expressed as uInplane ¼ uP ðuP en Þen and specimen. The value of the first-order diffracted angle of the
P
grating calculated using equation (2) was 3.99○. The dis-
uInplane
P ¼ uP ðuP en Þen . Combining these two tance between the sample and the grating was fixed at
expressions with equation (5) gives a relationship be- 28.7 mm, which insured that the diffracted images were
tween uInplane
P and uInplane
P , not overlapping. The rays passing from the specimen pro-
cessed by the grating were collected by a long distance
uInplane
Pþ ¼ uInplane ðuP en Þ tan θ e? ;
Inplane
P ð6a; bÞ microscope lens (K2/S, Infinity Photo-Optical Company,
uP ¼ uInplane
P þ ðuP en Þ tan θ e? ; Boulder, CO), and focused on the CMOS camera with a
758 Exp Mech (2013) 53:755–765
Transmission Pressure
grating chamber
(b) To pressure
Thin steel sensor Pressure
Face plate plates
Chamber
-1st order
Illumination
Camera Imaging lens 0th order
light
+1st order
Transmission
grating PDMS
Glass
membrane
To syringe window
(c)
PDMS
membrane
Transmission
grating
resolution of 2048 by 1536 pixels (3.2 μm pixel size) and microscope (TEM) test grid with square pattern (pitch size,
imaging speed of 12 frames per second (PL-B623, Pixe- 120 μm) are shown in Fig. 3(a), as seen in classic 3D-DIC.
LINK, Ottawa, Canada). A working distance of 152.4 mm Figure 3(b) shows three views of the same TEM square grid
between the sample and the front of the camera lens was obtained using the beam splitter grating, corresponding to
kept constant for all the experiments. The pixel size in the the negative first-order (−1), zeroth-order (0), and positive
sample plane was calibrated with a high precision stage first-order (+1) diffraction. The grating gives three undis-
micrometer to be 2.93 μm/pixel. torted face views of the grid, which are in contrast to the two
Generally, for classic 3D-DIC [8, 9], two images are distorted perspective views of the grid in Fig. 3(a).
obtained at different angles, one from Camera #1, and the The incident light rays from the specimen are processed
other from Camera #2, which result in two perspective through diffraction to create multiple images, resulting in a
views (distorted images). Two perspective views inclined more compact and simpler arrangement in comparison to
at ±40° of a circular region of a transmission electron traditional 3D-DIC techniques developed to date [8, 9, 15].
Exp Mech (2013) 53:755–765 759
Since the out-of-plane displacement of the specimen is first-order images, the proposed method eliminates the need
encoded in the in-plane displacements measured from the for a 3D-DIC algorithm. The full 3D displacement can be
(b)
760 Exp Mech (2013) 53:755–765
obtained with a 2D-DIC algorithm, which is relatively easier positive first-order diffraction, as shown in Fig. 4(a) and (b),
to implement and computationally much less intensive. were obtained before and after pressurization in the pressure
bulge experiment. These images were then processed with
Specimen and Loading Device the procedure described in Subsection 3.3. The correlation
results for the two sets of net full-field in-plane displace-
The proposed DAIC technique has been validated by conduct- ments, namely, (U +1, V +1) and (U −1, V −1) are shown in
ing pressure bulge and rigid-body translation and rotation Fig. 5. Note that the U +1 and U −1 fields contain both in-
experiments on a thin polymeric membrane made of polydi- plane and out-of-plane displacements of the membrane and
methylsiloxane (PDMS). The elastomer and curing agent of don’t exhibit any anti-symmetry about the central vertical
PDMS (Sylgard 184, Dow Corning Company, Midland, MI) axis. The components of the 3D deformation displacement
were mixed at a weight ratio of 10:1, and degassed in a field (u, v, w) are computed according to equation (8) by
vacuum chamber for 30 min to remove trapped air bubbles. using data from 2D-DIC, and are shown in Fig. 6(a)–(c).
The liquid PDMS premix was then spin-coated on a transpar- The computed out-of plane displacement field shows axial
ency film, followed by curing at 80 °C for 1 h. The cured thin symmetry corresponding to the deformation of the circular
membrane was peeled off from the transparency film and was membrane and the in-plane displacements also show appro-
cut to size for use in the experiments. For the pressure bulge priate symmetries and anti-symmetries, i.e., u (v) is sym-
experiment, the PDMS membrane was inflated up to a pres- metric about the central horizontal (vertical) axis and anti-
sure of 5 kPa. In order to create a random speckle pattern as symmetric about the central vertical (horizontal) axis.
well as the contrast which was required for DIC, the mem-
brane was covered with fine graphite powder (3–5 μm in size) Error Assessment
using a paint brush. For the rigid-body translation and rotation
experiments, spray painting was used to generate a speckle If the true displacement field of the inflated membrane is
pattern. The patterned membrane was clamped between two known, one can perform a quantitative error assessment by
thin steel disks, which were in turn attached to an airtight comparing the measured and true displacement fields of the
pressure chamber. In the pressure bulge experiment, the mem- membrane. In previous work employing bulge testing of a
brane specimen was inflated by increasing the pressure inside large membrane (a few to tens of centimeters in diameter)
the chamber with a syringe mounted on the optical table, and [15, 21–23], the true displacement used for comparison is
the pressure was monitored using a pressure sensor (GPS- well approximated with that obtained from either theoretical
BTA, Vernier, Beaverton, OR). The membrane had a thick- calculation or finite element modeling. However, the same
ness of approximately 40 μm, and a diameter of 1.9 mm for approximation is found to be very inaccurate in the present
the inflated area. work. The challenge here lies in the fact that the rigid-
clamping condition of the bulge test configuration tends to
Data Reduction be harder to satisfy as the size of the membrane becomes
smaller. Therefore, experiments involving rigid-body mo-
The images of the negative (−1) and positive (+1) first tion are chosen to perform quantitative error assessment
orders were acquired and processed using Vic-2D software from which the true displacements are easy to obtain.
(Correlated Solutions, Inc., W. Columbia, SC) to obtain the It is straightforward to estimate the error associated with
net in-plane displacements (U −1, V −1) and (U +1, V +1). The the proposed 3D displacement measurement scheme. The
correlation was performed between images obtained from theoretical derivation used in obtaining equation (8)
the same order of diffraction, i.e., between positive first- involves no approximations, and hence is strictly valid and
order (+1) images and between negative first-order (−1) does not introduce any systematic error. Neglecting any
images before and after deformation. The data was exported errors due to geometrical misalignment, if ε is the error
into MATLAB and the in-plane-displacements (u, v) as well present in the displacements (U +1, V +1) and (U −1, V −1),
as the out-of-plane displacement (w) were computed using the final measurement errors are εu,v 0ε in the measured x
equation (8) (a)–(c), respectively. and y displacements and "w ¼ "=tan θ in the z displacement
according to equation (8). In the present work, ε has two
possible sources for error: the first is inherent to 2D-DIC
Results measurement and the second is caused by image distortion
due to grating imperfection.
Validation of the Method The first type of error was evaluated by performing a
direct 2D-DIC measurement without using the transmission
Two sets of three speckle images of the circular membrane grating. Figure 7(a) shows the spatial distribution of mea-
corresponding to the negative first-order, zeroth-order and surement error of a constant x-displacement field, u06.03
Exp Mech (2013) 53:755–765 761
pixels (17.68 μm), which was produced by translating the (a) plots the contour map of the measured displacement, and
unpressurized membrane in the in-plane direction with a Fig. 8(b) shows a comparison between the measured and
high-precision translation stage. The sub-pixel error as true displacement profiles along the central horizontal axis.
shown is due to the combined effects of intensity interpola- The RMS error of the measured displacement was calculat-
tion and noise, lens aberrations, acoustic noise and vibra- ed to be 0.70 pixel, which is about 17 times the RMS error
tion. Subsequently, a second 2D-DIC measurement was (0.041 pixel) of in-plane displacement measurement. This
made on the same rigid-body translation by placing back ratio is in good accordance with the value of 1/tan3.99°0
the transmission grating and correlating two speckle images 14.3, thereby validating the expression for error estimation
of the negative first-order before and after translation. "w ¼ "=tan θ given earlier.
Figure 7(b) plots the error distribution of the new measure-
ment, which contains both the inherent 2D-DIC error and
the possible error originating from grating imperfection. The Discussion
image distortion effect due to the presence of the grating is
found to be negligible, since the root-mean-square (RMS) The expression for the out-of-plane error "w ¼ "=tan θ indi-
values of the two displacement error maps in Fig. 7(a) and cates that a large diffraction angle is necessary to achieve a
(b) are identical (0.041 pixel). The two error maps corre- small error in the out-of-plane (z) displacement measure-
spond to the same sensor plane location, enabling a fair ment. In the case of θ045°, the z displacement measurement
comparison by including the same amount of lens distortion. will have the same error level as the in-plane displacement
Additional rigid-body translation tests were also performed measurement. Higher measurement accuracy of the z dis-
for the y-displacement component and the positive first- placement may be attainable with θ larger than 45°. Never-
order, and led to the same conclusion. theless, as discussed in the Appendix, virtual objects formed
To assess the error in out-of-plane (z) displacement mea- at the positive and negative first orders of diffraction have a
surement, a linearly varying z displacement field along the non-zero image thickness. Increasing θ will increase this
x-direction was generated by rotating the unpressurized thickness and thus deteriorate the image quality if the thick-
membrane about a vertical axis for 11.0 degrees. Figure 8 ness becomes greater than the depth-of-field of the imaging
762 Exp Mech (2013) 53:755–765
(a)
0.15
0.1
-0.5
0.05
0 0
-0.05
0.5
-0.1
-0.15
-0.5 0 0.5
(b)
0.15
0.1
-0.5
0.05
0 0
-0.05
0.5 -0.1
-0.15
-0.5 0 0.5
(a) Pixel the distance between the grating and the measured object).
60 Therefore, simultaneous focus on the both images is diffi-
0.5 40 cult, especially at large diffraction angles. This fact limits
the use of blazed gratings to small diffraction angles.
20
y (mm)
(b)
Out-of-plane displacement (pixel)
80 Conclusions
Measured
60 True A new DIC-based experimental method, called the Diffrac-
40 tion Assisted Image Correlation (DAIC) method, has been
20
(a)
0
d
-20
d
-40
-60 e// A
-1 -0.5 0 0.5 1
en P
x (mm) e
B
Fig. 8 Measurement of an out-of-plane displacement field associated
with rigid-body rotation about an axis parallel to the y axis. (a) P+
Measured out-of-plane displacement field; (b) comparison between C
the measured and true displacement profiles along y00
Transmission
zeroth-order images does not include the out-of-plane dis- grating
placement, w, while the image correlation corresponding to
the first order contains the projected component of w. (b)
1.3
Hence, the following equations can be deduced to obtain
the 3D displacement components,
1.2
developed for 3D displacement measurement of small to Then, the grating equation sinðθ þ ΔθÞ sin Δa ¼
moderately sized objects (~100 microns to a few centi- l=p ¼ sin θ is invoked to obtain an expression for Δα.
meters). By utilizing a one-dimensional (line) transmission By substituting this expression into equation (A1) one gets,
grating to obtain multiple-angle views of a test specimen,
the proposed method provides a simple and yet effective tanðθ þ ΔθÞ tan θ
solution to 3D displacement characterization with only a d þ ðΔθÞ ¼ d : ðA2Þ
tan sin1 ½sinðθ þ ΔθÞ sin θ
single camera and 2D-DIC calculation. The validity and
accuracy of the method have been successfully demonstrat- The limiting value of d + as Δθ approaches zero can be
ed through 3D displacement measurement of a small soft obtained analytically. Denote the numerator in equation
membrane. This method eliminates the need for stereoscop- (A2) by f (Δθ), and the denominator by g(Δθ). Both f(Δθ)
ic imaging and/or use of multiple mirrors, thus resulting in a and g(Δθ) vanish as Δθ approaches zero. Using L’Hospi-
compact and simple setup. The use of 2D correlation and tal’s rule, in the limit as Δθ approaches zero,
algebraic operations to obtain full 3D displacements instead
of 3D correlation used in traditional methods significantly f 0 jΔθ!0 d
reduces the computational effort. With minimal effort, this d þ jθ!0 ¼ d ¼ : ðA3Þ
g jΔθ!0 cos3 θ
0
method can be implemented into microscopy systems for
deformation characterization of micro-scale objects, such as Figure 9(b) plots the curves of d + (normalized by
MEMS devices, micro/nano structured materials, and bio- d cos3 θ) versus Δθ for various values of θ. It is seen that
logical tissues and cells. d + varies around d cos3 θ with change in Δθ, implying a
non-zero thickness of the virtual image. d + is observed to be
Acknowledgments The research reported in this paper was con-
ducted while SX was a postdoctoral scholar and AG was a summer less sensitive to change in Δθ when θ is small.
undergraduate research fellow (SURF) in the Graduate Aerospace
Laboratories at the California Institute of Technology (GALCIT). It
was supported by the Caltech Center for the Predictive Modeling and References
Simulation of High-Energy Density Dynamic Response of Materials
through the U.S. Department of Energy’s National Nuclear Security
Administration Award# DE-FC52-08NA28613, which is gratefully 1. Peters WH, Ranson WF (1982) Digital imaging techniques in
acknowledged. The authors thank M. Mello and J. Notbohm for experimental stress analysis. Opt Eng 21:427–431
stimulating discussions. 2. Sutton MA, Wolters WJ, Peters WH, Ranson WF, McNeill SR
(1983) Determination of displacements using an improved digital
correlation method. Image Vision Comput 1:133–139
3. Chu TC, Ranson WF, Sutton WA, Peters WH (1985) Applications
Appendix: Location of a virtual image formed of digital-image-correlation techniques to experimental mechanics.
by diffraction of a transmission grating Exp Mech 25:232–244
4. Sutton MA, Li N, Joy DC, Reynolds AP, Li X (2007) Scanning
electron microscopy for quantitative small and large deformation
The ray diagram in Fig. 9(a) shows two selected rays measurements, Part I: SEM imaging at magnifications from 200 to
from a monochromatic point source of light, P, which is 10,000. Exp Mech 47:775–787
placed at a distance, d, from a transmission line grating. 5. Sutton MA, Orteu JJ, Schreier H (2009) Image correlation for
The definitions of the three unit vectors (e⊥, e//, en) are shape, motion and deformation measurements: basic concepts,
theory and applications. Springer, New York
the same as those in Fig. 1. The first-order diffraction 6. Choi S, Shah SP (1997) Measurement of deformations of concrete
angle of the grating is θ ¼ sin1 ðl=pÞ where l is the subjected to compression using image correlation. Exp Mech
wavelength of the light and p is the pitch of the grating. 37:307–331
7. Vendroux G, Knauss WG (1998) Submicron deformation field
The first ray (PB) impinges the grating at an angle of θ, measurements: part 2. Improved digital image correlation. Exp
and bends toward the normal to the grating surface after Mech 38:86–92
passing through the grating. The second incident ray 8. Luo PF, Chao YJ, Sutton MA, Peters WH (1993) Accurate mea-
(PC) deviates from the first by an angle of Δθ, resulting surement of three-dimensional deformations in deformable and
rigid bodies using computer vision. Exp Mech 33:123–132
in a different angle of diffraction, Δα. When one traces 9. Helm JD, McNeill SR, Sutton MA (1998) Improved three dimen-
back the diffracted rays, a virtual image is formed at P +, sional image correlation for surface displacement measurement.
where the two back-projected rays intersect. According to Opt Eng 35:1991–1920
the geometric relationship depicted in Fig. 9(a), the dis- 10. Pan B, Qian K, Xie H, Asundi A (2009) Two-dimensional digital
image correlation for in-plane displacement and strain measure-
tance between P + and the grating is ment: a review. Meas Sci Technol 20:062001
11. Orteu JJ (2009) 3-D computer vision in experimental mechanics.
Opt Laser Eng 47:282–291
jBC j tanðθ þ ΔθÞ tan θ
d þ ðΔθÞ ¼ ¼d : ðA1Þ 12. Tong W (2004) Plastic surface strain mapping of bent sheets by
tan Δa tan Δa image correlation. Exp Mech 44:502–511
Exp Mech (2013) 53:755–765 765
13. Tay CJ, Quan C, Huang YH, Fu Y (2005) Digital image correlation 19. Brauser S, Kromm A, Kannengiesser T, Rethmeier M (2010)
for whole field out-of-plane measurement using a single camera. In-situ synchrotron diffraction and digital image correlation tech-
Opt Commun 251:23–36 nique for characterizations of retained austenite stability in low-
14. Quan C, Tay CJ, Sun W, He X (2008) Determination of three alloyed transformation induced plasticity steel. Scripta Mater 63:
dimensional displacement using two-dimensional digital image 1149–1152
correlation. Appl Opt 47:583–593 20. Becker T, Splitthof K, Siebert T, Kletting P (2006) Error estima-
15. Pankow M, Justusson B, Waas AM (2010) Three-dimensional tions of 3D digital image correlation measurements. Proc SPIE
digital image correlation technique using single high-speed camera 6341:63410F
for measuring large out-of-plane displacements at high framing 21. Boyce BL, Grazier JM, Jones RE, Nguyen TD (2008) Full-field
rates. Appl Opt 49:3418–3427 deformation of bovine cornea under constrained inflation condi-
16. Dally JW, Riley WF (1991) Experimental stress analysis, 3rd edn. tions. Biomaterials 29:3896–3904
McGraw-Hill, New York 22. Meunier L, Chagnon G, Favier D, Orgéas L, Vacher P (2008)
17. Tippur HV, Krishnaswamy S, Rosakis AJ (1991) A coherent Mechanical experimental characterization and numerical model-
gradient sensor for crack tip deformation measurements: analysis ling of an unfilled silicone rubber. Polym Test 27:765–777
and experimental results. Int J Fract 48:193–204 23. Machado G, Favier D, Chagnon G (2011) Membrane curvatures
18. Hu Z, Xie H, Hua T, Wang Z (2009) Advanced intensity correla- and stress–strain full fields of axisymmetric bulge tests from 3D-
tion method for evaluating Poisson’s ratio of fiberlike material. DIC measurements. Theory and validation on virtual and experi-
Rev Sci Instrum 80:013105 mental results. Exp Mech 52:865–880