A fast data reduction algorithm for molecular tagging velocimetry: the decoupled spatial correlation technique

摘要

Molecular tagging velocimetry (MTV) involves intensive data reduction that extracts flow velocity information from the Lagrangian tracking of phosphorescing fluid material. A computationally efficient algorithm for the data reduction is thus of practical interest for processing large MTV data sets. We were motivated by this consideration into developing a simplified version of the spatial correlation technique, the decoupled spatial correlation technique, in an effort to seek a balance between accuracy and efficiency. By Taylor series analysis it is shown that, if the Lagrangian displacement vector can be roughly pre-determined, the two components in the displacement vectors that have to be solved simultaneously using the spatial correlation technique can now be determined independently in two orthogonal directions. This decoupling results in about an order of magnitude decrease in the CPU time required. An accuracy estimate based on artificial images that follow the motion of a line vortex indicates that the technique can determine displacements to within 0.08 pixel. This technique was also used to process MTV images acquired in a cross stream plane of the transverse jet. This flow is characterized by a large scale counter-rotating vortex pair (CVP). The velocity fields obtained clearly show the existence of this CVP, which provides further verification of this technique.