Abstract
This paper scrutinises the general suitability of image mapping for particle image velocimetry (PIV) applications. Image mapping can improve PIV measurement accuracy by eliminating overlap between the PIV interrogation windows and an interface, as illustrated by some examples in the literature. Image mapping transforms the PIV images using a curvilinear interface-fitted mesh prior to performing the PIV cross correlation. However, degrading effects due to particle image deformation and the Jacobian transformation inherent in the mapping along curvilinear grid lines have never been deeply investigated. Here, the implementation of image mapping from mesh generation to image resampling is presented in detail, and related error sources are analysed. Systematic comparison with standard PIV approaches shows that image mapping is effective only in a very limited set of flow conditions and geometries, and depends strongly on a priori knowledge of the boundary shape and streamlines. In particular, with strongly curved geometries or streamlines that are not parallel to the interface, the image-mapping approach is easily outperformed by more traditional image analysis methodologies invoking suitable spatial relocation of the obtained displacement vector.
Highlights
Particle image velocimetry (PIV) has become a well-established measurement technique to evaluate the underlying flow field from experimental flow images
The analysis of PIV images in the vicinity of surfaces is typically hampered by a reduction in seeding density and the presence of light reflections, rendering the cross-correlation process unreliable
With image mapping intensity distributions are subsequently transformed into the curvilinear coordinate system converting the interface boundary into a straight line such that it can be excluded from the interrogation process, in theory circumventing the associated error
Summary
Particle image velocimetry (PIV) has become a well-established measurement technique to evaluate the underlying flow field from experimental flow images. The aim of this paper is to fill this lack by documenting and investigating the individual stages in the image mapping process and study the generality of image mapping for PIV applications from the generation of the body-conforming mesh for a generic surface of interest, to the effects of the deformation on the cross-correlation map and related displacement measurements. The authors present a novel technique to minimise distortions in the cross-correlation maps due to deformed particle images caused by the inherent image transformations. Instead it is argued that more conventional image analysis methodologies are more reliable and robust
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
