Abstract of
THE ANALYSIS OF THE FREQUENCY DOMAIN AS A DECISIVE
FACTOR DURING THE EVALUATION PROCESS OF DPIV IMAGES CONTAINING DISTURBING AREAS
T. Scheffler, , H. Siekmann
Technical University of Berlin Sekr. K2, Hydraulic
Turbomachinery Straße des 17. Juni 135 10623 Berlin Germany
.
Digital-Particle-Image-Velocimetry (DPIV) evaluation algorithms
based on correlation techniques are usually designed to run over the whole image.
If a DPIV image contains disturbing structures and areas, incorrect velocity data
will occur at these places. A data validation should recognise these unwanted
data, or they will be hidden during the post-processing with the structures themselves.
A more elegant method is to avoid the appearance of incorrect data. An investigation
of a decisive factor to predict the sense of a local DPIV evaluation is presented.
In addition, this prediction is useful to change locally the evaluation process
from DPIV to Particle-Tracking-Velocimetry (PTV). This is illustrated on a DPIV
image of an experimental flow. The decisive factor deals with the analysis of
the grey-value-histogram of the frequency domain of the interrogation areas of
DPIV images using a rather small neural network. The use of a neural network does
not extend the evaluation time on the investigated DPIV images. On the contrary,
a decrease is observed. The analysis of the frequency domain is designed for single-frame
double-exposure DPIV images.
Return to Optical
Diagnostics in Engineering, Volume 3, Part 1 Contents