Two-dimensional on-line particle imaging velocimetry

A high-performance algorithm is described for the on-line computation of two-dimensional velocity maps using particle displacement imaging. It relies on the computation of direct correlations in a multiple-scale spatial hierarchy. Layered rejection criteria based on correlation quality and conditional averaging techniques are employed to achieve the robustness required for continuous, unsupervised operation. Implementations are described using a UNIX-based workstation and a DOS-based personal computer with a real-time image processing/correlator subsystem. The cooperation of the algorithm with other real-time visualization techniques is demonstrated using the example of colour-coded streak integration.The authors wish to thank Dr. Philippe Georis of the Université Libre de Bruxelles (ULB) for making available to us video sequences of his Marangoni convection experiment.