A modified Richardson–Lucy deconvolution for rapid reconstruction of light field μPIV

Light field micro-particle image velocimetry (LF-μPIV) can achieve the accurate measurement of the three-dimensional microflow through the high-resolution reconstruction algorithm of tracer particle distribution in measurement volume such as Richardson–Lucy deconvolution (RLD). However, RLD is time-...

Full description

Saved in:
Bibliographic Details
Published in:Experiments in fluids 2022-03, Vol.63 (3), Article 59
Main Authors: Gu, Mengtao, Li, Jian, Xu, Chuanlong
Format: Article
Language:English
Subjects:
Algorithms
Deconvolution
Engineering
Engineering Fluid Dynamics
Engineering Thermodynamics
Fluid- and Aerodynamics
Heat and Mass Transfer
Microchannels
Particle image velocimetry
Point spread functions
Reconstruction
Research Article
Tracer particles
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Light field micro-particle image velocimetry (LF-μPIV) can achieve the accurate measurement of the three-dimensional microflow through the high-resolution reconstruction algorithm of tracer particle distribution in measurement volume such as Richardson–Lucy deconvolution (RLD). However, RLD is time-consuming, which significantly lowers the efficiency of the particle distribution reconstruction in LF-μPIV. Considering the sparse characteristic of particle distribution in measurement volume of LF-μPIV, a modified Richardson–Lucy deconvolution (MRLD) is proposed to improve the reconstruction efficiency in this paper. The convolutions of the voxel matrixes in measurement volume and their point spread functions are replaced by the sum of the products of the non-zero voxels in measurement volume and their corresponding point spread function. This greatly eliminates the redundant calculations caused by the zero voxels in measurement volume, and thus effectively decreases the reconstruction time. Further, the applicable condition and appropriate threshold of MRLD are numerically analyzed and determined, and its performance is compared with that of RLD. Experiments are also performed in a rectangular microchannel to verify the feasibility of the proposed MRLD. Results demonstrated that MRLD is more than 2 times faster than RLD to reconstruct the particle distribution with identical reconstruction quality and measurement accuracy in measurement volume. Graphical abstract
ISSN:0723-4864
1432-1114
DOI:10.1007/s00348-022-03407-8