Self-correction of the optical distortion effect of thermal plumes in particle image velocimetry.

The article addresses the challenge of optical distortion in particle image velocimetry (PIV) caused by refractive index variations in thermally driven convection flows, particularly due to thermal plumes in viscous fluids. It introduces a novel composite image-processing algorithm called the shadow-affected PIV region filter (SA-PIVR filter), which combines conventional filters—including non-local means denoising and morphological opening—to identify and remove regions corrupted by optical distortion while preserving the true in-plane flow patterns. The filter’s effectiveness is validated through synthetic tests based on GPU-accelerated ray tracing and numerical simulations, as well as applied successfully to experimental PIV data from a high-Rayleigh-number thermal convection setup using viscous corn syrup. This method requires only existing PIV data, avoids computationally expensive ray tracing or large training datasets, and can serve as a baseline for future advanced correction techniques; additionally, the optical distortion patterns themselves provide useful information on plume morphology.