Neural optical flow for planar and stereo PIV

TL;DR

Neural optical flow (NOF) introduces a continuous neural-implicit approach for particle image velocimetry, improving accuracy, robustness, and enabling direct flow property inference from PIV images.

Contribution

The paper presents a novel neural-implicit framework for PIV that enhances data compression, flow analysis, and integrates physical constraints for improved flow measurement.

Findings

NOF outperforms state-of-the-art OF methods on synthetic and experimental datasets.

It enables direct pressure inference and enforces mass conservation in PIV analysis.

The approach is applicable to various flow measurement techniques beyond PIV.

Abstract

Neural optical flow (NOF) offers improved accuracy and robustness over existing OF methods for particle image velocimetry (PIV). Unlike other OF techniques, which rely on discrete displacement fields, NOF parameterizes the physical velocity field using a continuous neural-implicit representation. This formulation enables efficient data assimilation and ensures consistent regularization across views for stereo PIV. The neural-implicit architecture provides significant data compression and supports a space-time formulation, facilitating the analysis of both steady and unsteady flows. NOF incorporates a differentiable, nonlinear image-warping operator that relates particle motion to intensity changes between frames. Discrepancies between the advected intensity field and observed images form the data loss, while soft constraints, such as Navier-Stokes residuals, enhance accuracy and enable direct pressure inference from PIV images. Additionally, mass continuity can be imposed as a hard constraint for both 2D and 3D flows. Implicit regularization is achieved by tailoring the network's expressivity to match a target flow's spectral characteristics. Results from synthetic planar and stereo PIV datasets, as well as experimental planar data, demonstrate NOF's effectiveness compared to state-of-the-art wavelet-based OF and CC methods. Additionally, we highlight its potential broader applicability to techniques like background-oriented schlieren, molecular tagging velocimetry, and other advanced measurement systems.