Spectral random masking: a novel dynamic masking technique for PIV in multiphase flows

A novel masking technique for particle image velocimetry (PIV)-analysis of multiphase flows is presented. With this new approach, the velocity-field of an unmasked particle fraction (e.g., PIV-tracers) can be determined without the influence of a second (masked) particle fraction (e.g., bubbles or solid particles). Starting from a series of segmented grayscale images in which different particle fractions are determined for each frame, different masking techniques are compared. The problems caused by state of the art masking in case of dynamic masks (individual masks for each frame) are discussed. As a solution, the novel spectral random masking algorithm is presented which replaces masked regions in the images by randomized intensity patterns in order to hide them from subsequent PIV-analysis. The advantages over conventional masking techniques are demonstrated for a model experiment of crystallizing ammonium-chloride solution.