Flow-adaptive data validation scheme in PIV

A post-interrogation data validation technique is used to remove the spurious vectors in particle image velocimetry (PIV) results. The local-median method with constant user-adjustable thresholds usually works well when an appropriate threshold is set for a specified flow field. However, the selection of the appropriate threshold is not always easy since no guidance, such as the under-detected and over-detected percentages, is available to follow. In addition, a single constant threshold is generally not applicable in complicated flows such as inhomogeneous gradient flows or vortical flows. A flow-adaptive data validation (FADV) scheme is proposed in this study to avoid the selection of the appropriate thresholds for specified flow fields. Simulated non-uniform gradient flows and vortical flows with a noise distribution are added with simulated single or clusters of error vectors to evaluate the performance of the FADV scheme. Its performance is compared with the local-median method. The results show that the performance of the FADV scheme is superior to the local-median method, although the performance of the latter is optimized with the optimal threshold in each set of simulated conditions. The FADV scheme is also applied in a real measured turbulent swirling flow in gas cyclones.