Exp Fluids (2008) 45:549–556
DOI 10.1007/s00348-008-0521-5
RESEARCH ARTICLE
Volume self-calibration for 3D particle image velocimetry
B. Wieneke
Received: 13 December 2007 / Revised: 25 March 2008 / Accepted: 16 May 2008 / Published online: 13 June 2008
Ó Springer-Verlag 2008
Abstract Planar self-calibration methods have become
standard for stereo PIV to correct misalignments between
laser light sheet and calibration plane. Computing cross-
correlation between images from camera 1 and 2 taken at
the same time, non-zero disparity vectors indicate rota-
tional and translational misalignments relative to the
coordinate system defined by a calibration plate. This
approach works well for thin light sheets but fails for
extended volumes recorded in 3D-PTV or tomographic
PIV experiments. Here it is primarily necessary to correct
calibration errors leading to triangulation errors in 3D-PTV
or in degraded tomographic volume reconstruction.
Tomographic PIV requires calibration accuracies of a
fraction of a pixel throughout the complete volume, which
is difficult to achieve experimentally. A new volumetric
self-calibration technique has been developed based on the
computation of the 3D position of matching particles by
triangulation as in 3D-PTV. The residual triangulation
error (‘disparity’) is then used to correct the mapping
functions for all cameras. A statistical clustering method
suitable for dense particle images has been implemented to
find correct disparity map peaks from true particle matches.
Disparity maps from multiple recordings are summed for
better statistics. This self-calibration scheme has been
validated using several tomographic PIV experiments
improving the vector quality significantly. The relevance
for other 3D velocimetry methods is discussed.
B. Wieneke (&)
LaVision GmbH, Anna-Vandenhoeck-Ring 19,
37081 Gottingen, Germany
e-mail: bwieneke@lavision.de
1 Introduction
Standard in situ calibration methods for stereo PIV and
volumetric velocimetry techniques employ an accurately
manufactured planar calibration plate with known marks
translated in a controlled way through the illuminated sheet
or volume. With a dual-plane 3D calibration plate a single
recording is sufficient for a thin sheet eliminating the need
for a translation stage. Alternatively, one can use multiple
recordings of a planar calibration plate at arbitrary posi-
tions and process by camera pinhole model and bundle
adjustment fit, which is a common procedure in computer
vision. The outcome of the calibration is a mapping func-
tion M(X, Y, Z) from space to camera pixel coordinates
(x, y):
ðxi ; yi Þ ¼ Mi ðX; Y; Z Þ ð1Þ
for all cameras i. Details concerning calibration procedures
and functional forms of the mapping function can be found
in the literature (Soloff et al. 1997; Prasad 2000; Calluaud
and David 2004; Scarano et al. 2005). For stereo PIV there
are basically two types of calibration errors.
First the mapping function M itself can be erroneous due
to inaccurate calibration plates, inaccurate plate movement,
mechanical instabilities, optical distortions not accounted
for by the mapping function and other error sources. Even
with great care the maximum errors are typically above
0.5–2 pixel in some region of the mapped volume. This is
usually not a problem for stereo PIV, because the 2C-dis-
placement vectors are still computed accurately for both
cameras, only for the 3C-reconstruction the two vectors are
taken from slightly different world positions. Apart from
the case of very high velocity gradients a calibration error
of 1 pixel is not relevant compared to other common error
sources.
123
550
More significant for stereo PIV is the second type of
error resulting from a translational and rotational mis-
alignment between the coordinate system defined by the
calibration plate position (at z = 0) compared to the actual
light sheet position. Typical disparity errors are in the
range of 5–20 pixel. Advanced ‘self-calibration’ or ‘mis-
alignment correction’ methods have been developed
correcting such a misalignment using the actual recorded
particle images (Willert 1997; Coudert and Schon 2001;
Scarano et al. 2005; Wieneke 2005). This is done by taking
Exp Fluids (2008) 45:549–556
first summarized followed by a detailed explanation of
each step.
The first step is localizing 2D particle positions in the
image of each camera and finding corresponding 3D par-
ticle positions by matching and triangulation which is
essentially the first part of the 3D-PTV algorithm as shown
in Fig. 1 for three cameras. One should note that 3D-PTV
and tomographic PIV is usually done with four cameras.
The 2D-particle positions (x, y) of camera i of a single
particle in space are triangulated into a best-fit world
images from camera 1 and 2 recorded at the same time, position (X, Y, Z). The optimization criterion is  usually


then dewarping them to the world plane z = 0, which
corresponds to the position of the calibration plate, fol-
lowed by cross-correlation. The resulting 2D disparity
vector field should be zero, that is, the two images are
exactly matched. Non-zero disparity vectors can be used in
an appropriate way to shift and rotate the coordinate system
such that the new z = 0 plane coincides with the actual light
sheet position.
Even large displacements can be detected and corrected
eliminating the need to accurately align the calibration
plate with the light sheet in the first place. One should note
that the disparity correlation peak is actually an extended
streak with a length proportional to light sheet thickness.
The correction method works well for thin Gaussian light
sheets, but fails for thicker sheets and volumes since the
correlation streak becomes too long, too weak and frag-
mented, even showing multiple peaks from non-Gaussian
intensity profiles.
For scanning light sheet stereo PIV (Brücker 1996; Hori
and Sakakibara 2004) this self-calibration method can be
used to detect the exact position for all light sheets with a
single rough volume calibration before. Due to inaccura-
cies of the mechanical scanning mechanism such a
calculation might be required for each recorded image.
For 3D-PTV and especially for tomographic PIV with
extended volumes the first type of error relating to inac-
curate mapping functions is more relevant. The
tomographic reconstruction step requires that each voxel
position in space is mapped to a camera pixel position with
an error less than 0.4 pixel (Elsinga et al. 2006), preferably
less than 0.1 pixel. Therefore a volumetric self-calibration
technique is needed, which is capable of correcting the
mapping function throughout the volume. This is described
in the next section, followed by an experimental verifica-
tion using tomographic PIV showing significant
improvement of the vector field quality.
2 Method
The basic principle of correcting mapping function errors
in a volume using the actual recorded particle images is
123
done by minimizing the sum of the distances  x0i ; y0i

ðxi ; yi Þj where ðx0i ; y0i Þ is the projection of (X, Y, Z) back to
image of camera i (Hartley and Sturm 1994). For a perfect
mapping function the line of sights of all cameras intersect
in a single world point resulting in a zero disparity di of



di ¼ dix ; diy ¼ x0i ; y0i
ðxi ; yi Þ ð2Þ
Collecting disparities di for all particles throughout the
volume is then used to correct the mapping functions M to
M0
Mi0 ðX; Y; Z Þ ¼ Mi ðX; Y; Z Þ
di ðX; Y; Z Þ ð3Þ
with an appropriate validation and interpolation of the
sparse disparities di for each particle into a continuous field
or a similar method described below in more detail.
This basic principle works well with sparsely seeded
3D-PTV images with, for example, 1,000–5,000 particles
in each image, where the matching of corresponding par-
ticle images is done with a high confidence level ([95%)
(Maas 1996). But it does not work without further refine-
ment for densely seeded images from tomographic PIV.
Here the particle seeding density is much higher with
typically [50,000 particles in each image equivalent to
particle densities per pixel in the images of up to 0.1 ppp
(particles per pixel) similar to standard planar PIV. Ideally
Fig. 1 Residual triangulation disparities di ðX; Y; Z Þ ¼ ðxi ; yi Þ

ðx0i ; y0i Þ due to calibration errors
Exp Fluids (2008) 45:549–556
one would record first a number of images with less
seeding density, but this is often inconvenient or even not
possible.
Figure 2 shows the procedure to match particle images
corresponding to the same particle in space. For each
particle in the image of camera 1 one finds particles located
within a stripe defined by the uncertainty emax along the
length Lz which is the projection of the line of sight of
camera 1 through the illuminated volume from z1 to z2
viewed by camera 2 (‘epipolar line’). For the combination
of particle images for camera 1 and 2 a 3D particle position
(X, Y, Z) is computed by triangulation. Then for each
particle in the stripe it is checked if in the image of camera
3 and 4 there is a corresponding particle within ±emax
around the projection of (X, Y, Z) onto camera 3 and 4
image. The allowed uncertainty emax must be larger than
the largest expected calibration error. For 3D-PTV with
low seeding density and four cameras one rarely detects
any incorrect ghost particles. The ratio R of ghost particles
to true particles is a strong function of the seeding density
nppp and allowed error emax:
R ¼ 8  Lz  e5max  n3ppp ð4Þ
Typical conditions with, for example, Lz = 200 pixel,
nppp = 0.05 ppp and allowed error emax = 4 pixel lead to
R = 205 times more ghost particles than true ones. Obvi-
ously it is then very difficult to identify the true particles
and their disparity to be used for the mapping function
Fig. 2 Particle triangulation
procedure
551
correction. In addition given such high densities the
probability of overlapping particles is high which leads to
erroneous 2D and 3D particle positions and noisy
disparities.
The solution is to apply an appropriate clustering tech-
nique and using not all particles but only the brightest, for
example, 10%. Immediately the number of ghost particles
is reduced by nearly several orders of magnitude with the
added advantage that even in case of overlapping particles
the bright particles are much less influenced by another
overlapping weak particle. Usually image preprocessing
including global and local intensity renormalization is
applied to the four camera images to subtract background
intensities and to correct particle intensity differences due
to (e.g., forward–backward) laser scattering differences. A
single threshold for all camera images can then be used so
that the group of particles above a certain threshold is about
the same in all camera images.
The complete self-calibration procedure can be divided
into the following steps:
a. Measuring 2D particle positions in all camera images.
b. Determining possible 3D particle positions by
triangulation.
c. Subdividing the complete illuminated volume into nx 
ny  nz sub-volumes.
d. Plotting the disparities of all particles in a sub-volume
as a 2D disparity map.
123
552
e. Summing disparity maps for each sub-volume and
all cameras over many recordings to get better
statistics.
f. For each disparity map detecting highest disparity peak
which is the most probable disparity vector di(X,Y,Z).
g. Validating and optionally smoothing of the 3D2C
disparity vector field di.
h. Correcting mapping functions Mi according to Eq. 3
for all cameras i.
i. Repeating the complete procedure as a check and for
further improvement.
Step a As described above a suitable threshold is
selected to only use the brightest, for example, 10% of
all 2D particle images. Particle fitting itself is done in a
standard way with a 5 9 5 2D Gauss fit, reduced to a
3 9 3 Gauss fit if a neighboring particles gets too close,
that is, if an outer rim pixel is brighter than the inner
3 9 3 core, falling back to a dual x and y 1D Gauss fit
which always work. For isolated particles the position
accuracy is typically below 0.1 pixel depending on
optical conditions and particle size etc.
Step b The triangulation procedure computes all possible
particles in space with a triangulation uncertainty di
(Eq. 2) smaller than emax as shown in Fig. 2.
Step c The complete illuminated volume is subdivided
into nx  ny  nz sub-volumes (e.g., 5 9 5 9 5) in order to
collect sufficient 3D particles in each sub-volumes.
Step d For each sub-volume and all 3D particles within
the sub-volume a small 2D Gauss blob is plotted at the
disparity position (dix, diy) in a square of, for example,
50 9 50 pixels corresponding to the allowed maximum
error ±emax. For each camera i a separate disparity map
is generated.
The computation of the disparity di can be done in
slightly different ways. The first step is always to use the
particle positions (xi, yi) in the images of camera i to
compute a best-fit world position (X, Y, Z) by triangulation
(Hartley and Sturm 1994) as in step b together with the
residual triangulation error di as the difference between
(xi0 , yi0 ) = Mi (X, Y, Z) and (xi, yi) (Eq. 2). One has an extra
degree of freedom which allows one to use one camera as a
reference assuming that the mapping function for this
camera is perfect thus correcting only the other mapping
functions. This is done by taking Z from the original tri-
angulation and computing an Xref and Yref by inverting the
mapping function for the reference camera:

1
ðXref ; Yref Þ ¼ Mref ðxref ; yref ; Z Þ ð5Þ
and using Xref and Yref for computing (xi0 , yi0 ) for the other
cameras:


x0i ; y0i ¼ Mi ðXref ; Yref ; Z Þ ð6Þ
123
Exp Fluids (2008) 45:549–556
which is then used in Eq. 2. Of course, for the reference
camera the disparity is zero. Whether one should use a
reference camera deserves further investigation. Tomo-
graphic experiments so far have been mostly indifferent,
with a few working better with one or the other method.
Having collected possible 3D particles in space one now
needs to separate disparities from true accurate 2D/3D
particle positions from ghost particles and inaccurate ones
due to overlapping particles. There are many suitable
clustering techniques available. The one used here is based
on displaying the disparities for each sub-volume in a 2D
disparity map as a histogram with each measurement point
plotted as a small 2D Gauss blob. For example, the dis-
parity map might be of size of, for example, ±30 pixel
corresponding to a maximum allowed disparity emax of ±3
pixel, and each disparity di is plotted at position (dix, diy) as
a 2D Gauss blob with a constant width of 2 pixel (0.2 pixel
in units of disparity error).
The width of the Gauss curves provides sub-pixel
accuracy for the position of the highest disparity peak (step
f) similar to correlation peaks in standard PIV. It should be
made larger in case of few detected particles. The height of
the Gauss curve is set to 1 for all particles. Alternatively,
one could use the particle intensity of the 2D peak finder to
account for the light intensity variations across the volume.
Such a display of the disparity allows a good visual
inspection of the data quality. This is shown in Fig. 3 for a
real tomographic PIV experiment. One can guess that the
true disparity peak is at the top-right, but there are not yet
enough particles for good statistics.
Step e Disparity maps from multiple recordings are
summed for better statistics. Most often a single image
does not contain sufficient particles to add-up to a sharp
Fig. 3 Disparity map for a single sub-volume
Exp Fluids (2008) 45:549–556
unfragmented disparity peak, especially when choosing a
large number of sub-volumes. Figure 4 shows the
emerging disparity peak when adding disparity maps
from up to 16 images. The final outcome is one disparity
map for each sub-volume in each z-plane for all cameras.
Figure 5 shows the disparity maps for all sub-volumes for
one camera in a series of z-planes spaced 1 mm apart. The
off-center disparity peaks in the fully illuminated bottom
z-plane correspond to a calibration error of up to 2 pixel.
Step f The sub-pixel disparity peak location is deter-
mined, for example, by a 2D or dual 1D-Gauss fit and
corresponds to the most likely disparity vector di. The
disparity peak width is a good indication of the accuracy
of the 2D peak finder. For the experiment in Fig. 4, it is
only slightly larger than the selected Gauss width of 0.2
pixel, suggesting that the 2D peak finder accuracy is
around 0.1 pixel. In rare cases it has been observed that
the disparity peak is smeared due to error gradients within
a single (large) sub-volume. Usually, the calibration
errors are highly uniform across the complete volume.
Step g The outcome is a 3D2C disparity vector field for
each camera, one vector for each sub-volume. This vector
field is validated using a 3D universal outlier median filter
(Westerweel and Scarano 2005) over a region of, for
example, 3 9 3 9 3 or 5 9 5 9 5 vectors. One can store
secondary disparity peaks in step f for possible selection
by the median filter in case the highest peak is eliminated.
If the selected volume is larger, for example, in z-direction
than the actual illumination, then complete z-planes of
sub-volumes might show random outlier vectors.
The median filter with a kernel of 5 9 5 9 5 is robust
enough to throw out even complete z-planes of random
vectors. Note that a 3D median filter is considerably more
robust than a 2D filter due to the increased number of
neighbor vectors within some spatial domain. Most often it
is indicated to apply some additional 3 9 3 9 3 vector
smoothing to reduce the error inherent in the whole process
(Fig. 6). The true disparity field usually varies little across
the image. So it is possible to smooth the estimated dis-
parity field heavily to prevent errors propagating into an
erroneous new mapping function.
Fig. 4 Summing disparity maps over 1–16 recordings. Color scaling is always from minimum to maximum intensity
553
Choosing an initial volume larger than the illuminated
volume is a good method to actually measure the size of
the illuminated volume or, respectively, the common area
viewed by all cameras. Figure 5 shows the disparity maps
for consecutive z-planes on one side of the volume indi-
cating at what z-plane the illumination starts. The volume
is also slightly tilted with respect to the coordinate system
as evidence by the left to right decrease in the disparity
peak visibility.
Step h The mapping functions Mi are corrected for each
camera separately according to Eq. 3. Since the disparity
vector fields are given only on grid positions defined by
the center of possibly very few sub-volumes one needs
to interpolate the vector field to intermediate positions.
We choose to calculate new mapping functions M0 on the
z-planes defined by the sub-volumes. For each z-plane an
array of 20 9 20 points is taken for which the difference
between the original mapping function and the disparity
field is computed and a new M0 is fitted using third-order
polynomial functions in x and y (similar to Soloff et al.
1997).
Step i The sequence of steps a–h can be repeated after
the first self-calibration to check whether the disparity
peaks are finally positioned at the center position (0,0).
One can choose in subsequent iterations a smaller
maximum allowed disparity emax and a lower threshold
for more true particles and at the same time less ghost
particles. Usually, two or three iterations are sufficient to
reduce the mapping function errors down to below 0.1
pixel. At the same time the number of sub-volumes and
the selected volume can be adjusted. Often it is
advantageous to start with (e.g., 2 9 2 9 2) sub-
volumes to first correct roughly the mapping functions
and use finer divisions (e.g., 10 9 10 9 10) later for a
detailed local analysis of the 3D calibration errors.
3 Application to tomographic PIV experiments
This procedure is tested for a tomographic PIV experiment
in water observing a transitional wake behind a finite
123
554 Exp Fluids (2008) 45:549–556

b Fig. 5 Disparity maps for decreasing depth z from top (not illumi-
nated) to bottom (fully illuminated). Sub-volume size in xyz is about
8 9 8 9 1 mm

length, circular cylinder with 3D instabilities at Re = 540

(Michaelis et al. 2006) with a volume of 73 9 46 9 16 mm
and a final 3D3C vector field with 141 9 89 9 29 vectors.
It turns out that the initial mapping function had errors of
up to 2–3 pixels close to the cylinder. After self-calibration
the errors are reduced to below 0.1–0.2 pixel.
The volume self-calibration procedure improves signif-
icantly the contrast of the reconstructed volume and the
quality of the vector field in terms of higher correlation
values and improved signal-to-noise ratios. the raw vector
field without any vector validation is shown in Fig. 7. Self-
calibration eliminates most outliers. The signal-to-noise
ratio of the (3D) vector correlation maps is almost twice as
large as before. Figure 8 shows the complete volume
vector field before and after the calibration procedure after
vector post-processing.
An important aspect is that without such a volumetric
self-calibration procedure one is not even aware of how
large the calibration errors actually are. As it turned out,
even with great care during the calibration procedure there
was hardly any experiment where the calibration errors did
not exceed 1 or 2 pixels in some corner of the volume.
In one experiment viewing with cameras from opposite
side of the volume it was possible to correct calibration
errors in the range of 7 pixels arising due to a double-sided
calibration plate tilted slightly relative to the z-direction in
which it was moved. This resulted in the two cameras on
the back side being shifted relative to the front side
cameras.
Not knowing how large the error actually was one
started out with emax = 3 pixel, not finding any peaks in the
disparity maps. Then one had to gradually increase emax
and carefully adjusts the particle detection threshold to
minimize the ghost particles and processing time. With
emax = 10 (400 times as many ghost particles as for emax =
3) and only using 2 9 2 9 2 sub-volumes it was finally
possible to see a faint disparity peak among the vast
amount of ghost particles. Once this large error had been
corrected in a first step it was then easy to further reduce
the calibration errors down to 0.1 pixels with fine local
resolution possible due to otherwise high-contrast, low-
noise images with particle densities less than 0.05 ppp.

4 Relevance for other PIV techniques

For standard stereo PIV and scanning light sheet stereo PIV
(Brücker 1996; Hori and Sakakibara 2004) the planar
stereo PIV self-calibration method is usually sufficient to

123
Exp Fluids (2008) 45:549–556
Fig. 6 Original disparity vector
field (left) and after vector
validation and smoothing (right)
measure and correct the position and rotation of the light
sheets which typically lead to disparities of 5–20 pixel.
Remaining calibration errors of the order of a few pixels
are only relevant in case of very strong local velocity
gradients. But especially for thick light sheets required, for
example, for cross-flow measurements the volumetric self-
calibration technique is recommended. Here standard pla-
nar self-calibration often fails if the light sheet profile is not
nicely Gaussian, but, for example, shows multiple intensity
peaks within the long correlation streak.
Thick light sheet PIV images can be processed with the
tomographic PIV algorithm usually adding a third or
fourth camera (Wieneke and Taylor 2006). Volumetric
Fig. 7 Z-plane of volume vector field without (top) and with (bottom)
volume self-calibration. Shown is the raw vector field without vector
validation with the 3D vorticity magnitude as background color and
every fourth vector in x-direction
555
self-calibration is then required to reduce the calibration
errors to acceptable levels.
Finally, volumetric self-calibration is also recommended
for standard 3D-PTV. For low seeding densities and unique
conversion of 2D particle positions into a 3D particle
position, a high systematic triangulation error is not criti-
cal. But for higher seeding densities the correction of
calibration errors by volumetric self-calibration makes it
possible to choose a very small threshold for the maximum
allowed triangulation error. This significantly reduces the
number of ghost particles and can provide some rejection
criteria for overlapping particles which otherwise lead to
inaccurate 3D particle positions.
Table 1 summarizes the importance of the two self-
calibration algorithms for correcting laser sheet misalign-
ments and mapping function errors, respectively, for the
Fig. 8 Complete 3D flow field without (top) and after self-calibration
(bottom) with 3D vorticity colored according to dominant sense of
rotation (Michaelis et al. 2006)
123
556
Table 1 Relevance of calibration errors for different techniques
Technique Inaccurate Misalignment
mapping between
function calibration
plate and
light sheet
Stereo PIV - ++
Thick-sheet tomographic PIV ++ ++
Full-volume tomographic PIV ++ - References
3D-PTV + -
+ Strong, - Weak
different planar and volumetric particle image velocimetry
techniques.
5 Summary
A volumetric self-calibration technique has been developed
to correct mapping function errors using the actual
recordings of the particle images. This is especially rele-
vant for tomographic PIV where the mapping errors should
be below 0.3 pixel. The method is based on detection of 2D
particle images and triangulation of the 3D particle position
in space. Particle images are usually intensity normalized
across the image and between cameras to account for
Mie-scattering differences, and a threshold is used to select
only the, for example, 10% brightest particles. Back-pro-
jecting the 3D position toward the camera images and
comparing to the original position provides a disparity
error for each particle. Collecting disparities for all parti-
cles throughout the volume and a suitable clustering
method to eliminate false 3D particles one obtains a 3D2C
disparity vector field for each camera. Sufficient statistics is
obtained by summing the disparities from many recordings.
After vector field validation and smoothing this disparity
field is then used to correct the original mapping function.
It is also possible to measure the actual extent and rotation
of the illuminated volume from the disparity maps.
The technique has been applied successfully to a number
of tomographic PIV experiments improving significantly
the quality of the volumetric vector field. It was found that
in most experiments even with a careful calibration pro-
cedure calibration errors of 1–2 pixel were found
somewhere in the volume. After self-calibration this has
123
Exp Fluids (2008) 45:549–556
been reduced to 0.1–0.2 pixel. The method is also useful
for stereo PIV especially with thicker light sheets since
here the standard planar self-calibration technique often
fails due to multiple intensity peaks within the light sheet.
For 3D-PTV the method can reduce the number of ghost
particles allowing higher seeding densities.
Brücker Ch (1996) 3-D scanning-particle-image-velocimetry: tech-
nique and application to a spherical cap wake flow. Exp Fluids
56:157–179
Calluaud D, David L (2004) Stereoscopic particle image velocimetry
measurements of a flow around a surface-mounted block. Exp
Fluids 36:33–61
Coudert S, Schon JP (2001) Back-projection algorithm with mis-
alignment corrections for 2D3C stereoscopic PIV. Meas Sci
Technol 12:1371–1381
Elsinga GE, Scarano F, Wieneke B, van Oudheusden BW (2006)
Tomographic particle image velocimetry. Exp Fluids 41:933–
947
Hartley R, Sturm P (1994) Triangulation. ARPA Image Understand-
ing Workshop, Monterey, pp 957–966
Hori T, Sakakibara J (2004) High speed scanning stereoscopic PIV
for 3D vorticity measurement in liquids. Meas Sci Technol
15:1067–1078
Maas HG (1996) Contributions of digital photogrammatry to 3D PTV.
In: Dracos T (ed) Three-dimensional velocity and vorticity
measuring and image annalysis techniques. Kluwer, Dordrecht,
pp 191–208
Michaelis D, Poelma C, Scarano F, Westerweel J, Wieneke B (2006)
A 3D time-resolved cylinder wake survey by Tomographic PIV.
In: 12th international symposium on flow visualization, Göttin-
gen, Germany, 10–14 September
Prasad A (2000) Stereoscopic particle image velocimetry. Exp Fluids
29:103–116
Scarano F, David L, Bsibsi M, Calluaud D (2005) S-PIV comparative
assessment: image dewarping + misalignment correction and
pinhole + geometric back projection. Exp Fluids 39:257–266
Soloff SM, Adrian RJ, Liu ZC (1997) Distortion compensation for
generalized stereoscopic particle image velocimetry. Meas Sci
Technol 8:1441–1454
Westerweel J, Scarano F (2005) A universal detection criterion for the
median test. Exp Fluids 39–6, 1096
Wieneke B (2005) Stereo-PIV using self-calibration on particle
images. Exp Fluids 39:267–280
Wieneke B, Taylor S (2006) Fat-sheet PIV with computation of full
3D-strain tensor using tomographic reconstruction. In: 13th
international symposium on applications of laser techniques to
fluid mechanics, Lisbon
Willert C (1997) Stereoscopic digital particle image velocimetry for
application in wind tunnel flows. Meas Sci Technol 8:1465–1469