particles. A 3D PTV system consists of the following
components:
* Image acquisition system with up to four CCD cameras
including a data storage device.
* Illumination facility.
e Tracer particles to seed and visualize the flow.
Whether high-grade components or off-the-shelf products come
into operation is depending on the experiment requirements as
well as on the budget. The data acquisition system defines the
spatial and temporal resolution.
The system used at the ETH Zurich was upgraded from offline
to online image digitization. In the previous system, the image
sequences were firstly recorded on analogue videotapes and
digitized afterwards, while in the new system two frame
grabbers (Matrox Genesis) are used to provide online
digitization and storage. The length of the recorded digital
image sequences is nowadays restricted by the storage device
capabilities. The data rate for a 60 Hz full-frame camera with a
resolution of 640 x 480 pixels is about 19 MB/sec, and hence in
an experiment which lasts for 1 minute four cameras deliver a
total amount of about 4.5 GB image data. Hardware setups for
3D PTV used at ETH Zurich are described in (Maas, 1992;
Virant, 1996; Stüer, 1999). Figure 1 shows a hardware setup
with four cameras used for the observation of the flow in a
aqueous copper sulfate fluid between two electrodes (Lüthi et
al, 2001).
Figure 1: Experiment setup with four CCD-cameras
3. SYSTEM CALIBRATION AND DATA ACQUISITION
A calibration of the system is required to be able to establish
correspondences between particle images and to compute 3D
particle coordinates. Therefore a reference body with discrete
targets is inserted into the observation volume before or after
the experiment. The images of the reference body are used to
determine the interior and exterior orientations as well as the
additional parameters of the cameras. Using the mathematical
model of spatial resection, the orientation and calibration
parameters of a camera can be determined from one single
image of the calibration reference body under suitable
illumination, if the 3D coordinates of the targets are known.
During the experiment synchronous image sequences of the
particle seeded flow are acquired. The particles appear as bright
spots that can be detected automatically from the images. An
example of a particle image used for 3D PTV is shown in
Figure 2.
Figure 2: Particle image for 3D PTV
4. PROCESSING STEPS
After the acquisition of the image sequences the data is
processed in the following steps:
e Image preprocessing by highpass filtering to remove non-
uniformities in the background illumination.
* Detection of particles in the images.
* Establishment of stereoscopic correspondences using
epipolar constraints. For details see (Maas, 1992; Dold and
Maas, 1994).
* Determination of 3D particle coordinates.
e Tracking in object- and image space.
The results of the automated processing are trajectories of the
particles in the three-dimensional velocity field. The processing
scheme of 3D PTV is shown in Figure 3.
Flow visualization
Image acquisition
Highpass filtering ||«& — |
a7 Image
Detection and preprocessing
location of particles
Y
Establishment of
_ correspondences Le
Determination of Camera orientations
3D-coordinates Calibration data
Ls | s auprauon data |
Add N°.
Tracking in object T
and image space Kinematic model
|
Particle trajectories
Figure 3: Processing scheme.
Additional particle positions are calculated during the tracking
—602—
A parti
path in
particle
particle
assignr
reconst
object
space r
Ob
Sp
The £
for tl
devel
spatic
corre!
imag;
ande
The
whos
in a
Lagr
cons
direc
calcı
to ge
The
estal
betv
mul
exis
