A Gaussian Smoothing-based velocity track denoising approach for 3D-PTV measurements

Volumetric particle tracking velocimetry (3D-PTV) is a well-established optical measurement technique that analyses complex flow structures by tracking the motion of several neutrally buoyant particles seeded in the fluid. Novel particle tracking methods such as Shake-the-Box (STB) have expanded the utility and accuracy of PTV. However, the obtained tracks have measurement noise, owing to several error sources present in the measurement chain. This adversely affects the accuracy of the reconstructed velocity field and other derived quantities. This work introduces a novel Gaussian kernel-based track smoothing approach for denoising particle tracks. The uniqueness of this technique lies in its ability to provide smoothened velocity tracks without any differentiation scheme on the position tracks. The Gaussian smoothing (GS) method was tested on synthetic cases of 3D Hama flow and Stokes vortex. GS reduced the root-mean-square-errors in the velocity and turbulent statistics by around 90% compared to standard PTV and 50% compared to FlowFit. Additionally, the effect of kernel width and length is investigated for optimal noise reduction. GS will be tested on experimental datasets and compared against the state-of-the-art methods (TrackFit, FlowFit) and other commonly used filters.