pith. sign in

arxiv: 2509.25924 · v4 · pith:GIQPF7Z2new · submitted 2025-09-30 · ⚛️ physics.flu-dyn

High Resolution and High-Speed Live Optical Flow Velocimetry

Juan Pimienta , Jean-Luc Aider This is my paper

Pith reviewed 2026-05-22 12:50 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn
keywords optical flow velocimetryparticle image velocimetryreal-time processingdense velocity fieldshigh-speed imagingfluid dynamicsGPU optimizationflow control
0
0 comments X

The pith

Optical flow velocimetry now computes one velocity vector per pixel in real time from particle images up to 1400 Hz for 1-megapixel frames.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents an optimized version of optical flow velocimetry that processes large particle image fields live while maintaining dense spatial coverage. It shows that this per-pixel approach can capture strong displacement gradients in benchmark flows such as a Rankine vortex and isotropic turbulence when seeding is adequate. The resulting live velocity data support immediate calculation of derived flow quantities over extended periods. These capabilities open the way to in-experiment monitoring and real-time feedback in fluid experiments.

Core claim

Through targeted algorithmic refinements and GPU optimizations, optical flow velocimetry achieves real-time dense velocity fields with one vector per pixel at frequencies reaching the kilohertz range, specifically handling 32-megapixel images at 90 Hz, 4-megapixel images at 460 Hz, and 1-megapixel images at 1400 Hz, as confirmed on synthetic Rankine vortex and turbulence cases plus experimental cylinder wake measurements.

What carries the argument

The per-pixel optical flow computation, refined for speed and paired with GPU acceleration, replaces window-based correlation to produce dense fields directly from consecutive particle images.

If this is right

  • Dense instantaneous velocity fields become available during the experiment rather than only afterward.
  • Derived quantities such as vorticity can be computed and displayed live from the velocity data.
  • Sustained high-rate acquisition allows recovery of low-frequency flow dynamics over long durations.
  • Closed-loop flow control strategies can use the real-time OFV measurements directly.
  • Conventional post-processing of large image sets becomes faster and less computationally expensive.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The live dense fields could feed directly into adaptive experimental setups that adjust forcing in response to measured flow features.
  • Similar optimization steps might extend real-time dense imaging to other scalar or vector field measurements in fluids.
  • High-rate dense data could reduce the total number of separate experiments needed to characterize unsteady flows.

Load-bearing premise

The resolution of small-scale displacement gradients depends on achieving suitable and sufficient particle seeding in the flow.

What would settle it

A well-seeded synthetic Rankine vortex test case that still fails to resolve known strong gradients at small scales would falsify the resolution performance claim.

Figures

Figures reproduced from arXiv: 2509.25924 by Jean-Luc Aider, Juan Pimienta.

Figure 1
Figure 1. Figure 1: FIG. 1: Diagram showing the main steps used to compute the velocity vectors at the [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2: Theoretical displacement magnitude of the Rankine vortex for various radii and [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3: Theoretical velocity profiles used to move the particles in the images. (a) Velocity [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: FIG. 4: 3D representation of the selected plane from a velocity datacube snapshot from [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: FIG. 5: Sample of DNS velocity fields. (a) [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: FIG. 6: Synthetic particle images for the HIT dataset. Left to right: increasing particle [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: FIG. 7: a) Principle used to obtain all theoretical displacement profiles for the Rankine [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: FIG. 8: Theoretical vs OFV displacement profiles for a Rankine vortex ( [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: FIG. 9: Scatter-plot error analysis for a Rankine vortex with core radius [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: FIG. 10: Best OFV results for each case. The core radius of the vortex changes along the [PITH_FULL_IMAGE:figures/full_fig_p018_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: FIG. 11 [PITH_FULL_IMAGE:figures/full_fig_p019_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: FIG. 12: Rankine vortex with [PITH_FULL_IMAGE:figures/full_fig_p020_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: FIG. 13: Rankine vortex with [PITH_FULL_IMAGE:figures/full_fig_p021_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: FIG. 14: Absolute displacement-error maps [PITH_FULL_IMAGE:figures/full_fig_p022_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: FIG. 15: Spatially averaged absolute displacement error [PITH_FULL_IMAGE:figures/full_fig_p023_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: FIG. 16: Snapshots of velocity magnitude: [PITH_FULL_IMAGE:figures/full_fig_p024_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: FIG. 17: Comparison between DNS, OFV, and CC-PIV along a mid-height profile. [PITH_FULL_IMAGE:figures/full_fig_p025_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: FIG. 18: Spatially averaged kinetic energy over the 199 time steps. Comparison between [PITH_FULL_IMAGE:figures/full_fig_p025_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: FIG. 19: Spectra averaged over 199 time steps, computed from the mid-height kinetic [PITH_FULL_IMAGE:figures/full_fig_p026_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: FIG. 20: Structure detection — DNS [PITH_FULL_IMAGE:figures/full_fig_p026_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: FIG. 21: Structure detection - OFV. a) Full field, red square marks a region of interest. b) [PITH_FULL_IMAGE:figures/full_fig_p027_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: FIG. 22: Structure detection - CCPIV. a) Full field, red square marks a region of interest. [PITH_FULL_IMAGE:figures/full_fig_p028_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: FIG. 23: Post-processing performance showing the number of velocity fields computed per [PITH_FULL_IMAGE:figures/full_fig_p029_23.png] view at source ↗
Figure 24
Figure 24. Figure 24: shows the number of velocity fields computed per second during the experiment as a function of the size of the images. The computing speed also scales linearly with the image size, similarly to the off-line case. Of course the performances also depends on the chosen OFV parameters, especially the number of PSL. Despite the impact of the data transfer from the camera to the GPU, the computing speeds measur… view at source ↗
read the original abstract

Particle Image Velocimetry (PIV) typically relies on cross-correlation,which makes it difficult to obtain instantaneous velocity fields that are both spatially dense and available in real time at high acquisition rates. Optical Flow Velocimetry (OFV) offers a per-pixel alternative. Here we demonstrate real-tome OFV that delivers dense velocity fields (one vector per pixel) with high effective spatial resolution at frequencies up to the kHz range. Using synthetic particle images for two benchmarks -- a Rankine vortex and a homogeneous isotropic turbulence DNS -- we show that, with suitable particle seeding, OFV can resolve strong displacement gradients down to small scales. We then achieve real-time performance through algorithmic refinements and GPU-focused optimizations, combined with practical choices of OFV parameters. With this implementation, 32 Mp fields are processed live at 90 Hz, 4 Mp fields up to 460 Hz, and 1 Mp fields up to 1400 Hz. The method is further validated experimentally on the flow past a circular cylinder, where dense instantaneous velocity fields support real-time computation of derived quantities over long durations. These capabilities enable in-experiment monitoring, recovery of low-frequency dynamics from sustained high-rate acquisition, and closed-loop-flow-control strategies based on OFV measurements while also accelerating conventional post-processing to reduce turnaround time and computational cost.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript presents an optimized real-time Optical Flow Velocimetry (OFV) implementation that produces dense, per-pixel velocity fields from particle images. It reports processing rates of 90 Hz for 32 Mp fields, 460 Hz for 4 Mp fields, and 1400 Hz for 1 Mp fields, achieved via algorithmic refinements and GPU optimizations. Validation is provided on synthetic benchmarks (Rankine vortex and homogeneous isotropic turbulence DNS) showing resolution of strong displacement gradients under suitable seeding, plus experimental demonstration on the wake of a circular cylinder enabling real-time computation of derived quantities.

Significance. If the reported performance and resolution hold under typical experimental conditions, the work would enable live high-speed dense velocimetry for in-experiment monitoring, recovery of low-frequency dynamics from sustained acquisition, and closed-loop flow control, while also accelerating post-processing relative to cross-correlation PIV. The explicit GPU-focused optimizations and concrete frame-rate numbers for different resolutions constitute a practical contribution.

major comments (1)
  1. [§3] §3 (synthetic benchmarks): The claim that OFV resolves strong displacement gradients down to small scales is demonstrated only at a single nominal seeding level for both the Rankine vortex and HIT DNS cases. No sensitivity study is reported that varies particle density, diameter, or out-of-plane motion while quantifying the smallest resolved scale or gradient error; this leaves the robustness of the high effective spatial resolution claim dependent on an unquantified 'suitable' seeding window.
minor comments (2)
  1. [Abstract] Abstract: 'real-tome' is a typographical error and should read 'real-time'.
  2. [Abstract] Abstract and validation sections: The central performance claims would be strengthened by inclusion of quantitative error bars, direct comparison tables against PIV or other OF methods, and full parameter settings for the reported benchmarks.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript. We address the major comment point by point below and indicate where revisions will be made to strengthen the work.

read point-by-point responses

  1. Referee: [§3] §3 (synthetic benchmarks): The claim that OFV resolves strong displacement gradients down to small scales is demonstrated only at a single nominal seeding level for both the Rankine vortex and HIT DNS cases. No sensitivity study is reported that varies particle density, diameter, or out-of-plane motion while quantifying the smallest resolved scale or gradient error; this leaves the robustness of the high effective spatial resolution claim dependent on an unquantified 'suitable' seeding window.

    Authors: We agree that the original presentation relied on a single nominal seeding level and that an explicit sensitivity study would better quantify the robustness of the high-resolution claim. In the revised manuscript we will add a new sensitivity analysis in §3 that systematically varies particle seeding density and diameter for both the Rankine vortex and HIT DNS cases. We will report the resulting changes in the smallest reliably resolved scale and in gradient error, thereby defining a practical 'suitable' seeding window. Out-of-plane motion is not varied in the synthetic benchmarks because those cases are strictly two-dimensional; however, the experimental cylinder-wake data already incorporate realistic out-of-plane effects, and the method continues to deliver usable fields. These additions will be presented without altering the reported real-time performance figures. revision: yes

Circularity Check

0 steps flagged

No significant circularity; claims rest on measured benchmarks and optimizations

full rationale

The paper reports measured real-time frame rates (32 Mp at 90 Hz, etc.) achieved via algorithmic refinements, GPU optimizations, and parameter choices, validated against synthetic Rankine vortex and HIT DNS cases plus cylinder experiments. These are external performance numbers tied to hardware and seeding conditions rather than any derivation that reduces by construction to fitted inputs, self-definitions, or self-citation chains. No equations or uniqueness theorems are invoked that loop back to the target results; the derivation chain is self-contained against the stated benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard optical flow assumptions plus practical parameter choices for real-time execution; no new physical entities are postulated.

free parameters (1)
  • OFV parameters
    Practical choices of OFV parameters are invoked to achieve the stated real-time rates.
axioms (1)
  • domain assumption Optical flow assumptions hold for suitably seeded particle images in fluid flows
    Invoked when claiming resolution of strong displacement gradients on synthetic benchmarks.

pith-pipeline@v0.9.0 · 5762 in / 1260 out tokens · 47108 ms · 2026-05-22T12:50:47.715133+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Real-Time Estimation of High-Resolution Flow Fields and Reduced-Order Coordinates from Event-Based Imaging Velocimetry

    physics.flu-dyn 2026-05 unverdicted novelty 6.0

    A framework uses offline-paired LR/HR data and POD latent-space linear models with Kalman filtering to reconstruct high-resolution velocity fields from coarse real-time event-based velocimetry, outperforming cubic int...

Reference graph

Works this paper leans on

58 extracted references · 58 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    To solve Eq

    imposes that neighboring pixels exhibit similar motion. To solve Eq. 1 an iterative min- imization by sum of squared differences is used. Algorithmically, the LK algorithm is a compromise between purely differential optical flow and window-based approaches: it uses a pixel-centered kernel, whose kernel radius (KR) sets the neighborhood over which intensit...

    work page

  2. [2]

    Local intensity normalization

    work page

  3. [3]

    Gaussian-pyramid subsampling (construction of coarse-to-fine levels)

    work page

  4. [4]

    Per-pixel displacement estimation at the kernel scale (LK with Gauss–Newton)

    work page

  5. [5]

    Upsampling and projection of the displacement to the next finer level

    work page

  6. [6]

    Iteration of steps 3–4 at each level until convergence

    work page

  7. [7]

    convergence zone

    Final dense velocity-field computation at full resolution. FIG. 1: Diagram showing the main steps used to compute the velocity vectors at the kernel scale. The kernel-scale estimation is sketched in Fig. 1. Two subimages of size 12 ×12 px2 at times t and t′= t + δtare considered. In this example, two neighboring pixels at 6 (x,y ) = (6, 6) (red) and (7, 6...

    work page 2023

  8. [8]

    Raffel, C

    M. Raffel, C. E. Willert, F. Scarano, C. J. K¨ ahler, S. T. Wereley, and J. Kompenhans,Particle image velocimetry: a practical guide (springer, 2018)

    work page 2018

  9. [9]

    Theunissen, F

    R. Theunissen, F. Scarano, and M. L. Riethmuller, An adaptive sampling and windowing interrogation method in piv, Measurement Science and Technology 18, 275 (2006)

    work page 2006

  10. [10]

    Scarano and M

    F. Scarano and M. L. Riethmuller, Advances in iterative multigrid piv image processing, Experiments in fluids 29, S051 (2000)

    work page 2000

  11. [11]

    Keane, R

    R. Keane, R. Adrian, and Y. Zhang, Super-resolution particle imaging velocimetry, Measure- ment Science and Technology 6, 754 (1995)

    work page 1995

  12. [12]

    Susset, J

    A. Susset, J. Most, and D. Honor´ e, A novel architecture for a super-resolution piv algorithm developed for the improvement of the resolution of large velocity gradient measurements, Experiments in Fluids 40, 70 (2006)

    work page 2006

  13. [13]

    Stitou and M

    A. Stitou and M. Riethmuller, Extension of piv to super resolution using ptv, Measurement Science and Technology 12, 1398 (2001)

    work page 2001

  14. [14]

    Zhang and M

    M. Zhang and M. D. Piggott, Unsupervised learning of particle image velocimetry, in High Performance Computing: ISC High Performance 2020 International Workshops, Frankfurt, Germany, June 21–25, 2020, Revised Selected Papers 35 (Springer, 2020) pp. 102–115

    work page 2020

  15. [15]

    J. S. Choi, E. S. Kim, and J. H. Seong, Deep learning-based spatial refinement method for robust high-resolution piv analysis, Experiments in Fluids 64, 45 (2023)

    work page 2023

  16. [16]

    E. Ilg, N. Mayer, T. Saikia, M. Keuper, A. Dosovitskiy, and T. Brox, Flownet 2.0: Evolution of optical flow estimation with deep networks, inProceedings of the IEEE conference on computer vision and pattern recognition (2017) pp. 2462–2470

    work page 2017

  17. [17]

    Maruyama, Y

    T. Maruyama, Y. Yamaguchi, and A. Kawase, An approach to real-time visualization of piv method with fpga, in International Conference on Field-Programmable Logic and Applications (2001)

    work page 2001

  18. [18]

    Fujiwara, K

    T. Fujiwara, K. Fujimoto, and T. Maruyama, A real-time visualization system for piv, in International Conference on Field-Programmable Logic and Applications (2003)

    work page 2003

  19. [19]

    Siegel, K

    S. Siegel, K. Cohen, T. McLaughlin, and J. H. Myatt, Real-time particle image velocimetry for closed-loop flow control studies (2003)

    work page 2003

  20. [20]

    V. A. Shchapov, A. Pavlinov, E. N. Popova, A. Sukhanovskii, S. L. Kalyulin, and V. Y. Modor- 34 skii, Supercomputer real-time experimental data processing: Technology and applications, in Russian Supercomputing Days (2018)

    work page 2018

  21. [21]

    C. E. Willert, M. Munson, and M. Gharib, Real-time particle image velocimetry for closed-loop flow control applications (2010)

    work page 2010

  22. [22]

    McCormick, B

    F. McCormick, B. Gibeau, and S. Ghaemi, Reactive control of velocity fluctuations using an active deformable surface and real-time piv, Journal of Fluid Mechanics 985 (2024)

    work page 2024

  23. [23]

    N. A. Piga, Y. Onyshchuk, G. Pasquale, U. Pattacini, and L. Natale, Roft: Real-time optical flow-aided 6d object pose and velocity tracking, IEEE Robotics and Automation Letters 7, 159 (2021)

    work page 2021

  24. [24]

    Y. Wang, Z. Hou, K. Leman, and R. Chang, Real-time video stabilization for unmanned aerial vehicles, in IAPR International Workshop on Machine Vision Applications (2011)

    work page 2011

  25. [25]

    A. Lim, B. Ramesh, Y. Yang, C. Xiang, Z. Gao, and F. Lin, Real-time optical flow-based video stabilization for unmanned aerial vehicles, Journal of Real-Time Image Processing 16, 1975 (2017)

    work page 1975

  26. [26]

    Brebion, J

    V. Brebion, J. Moreau, and F. Davoine, Real-time optical flow for vehicular perception with low- and high-resolution event cameras, IEEE Transactions on Intelligent Transportation Sys- tems 23, 15066 (2021)

    work page 2021

  27. [27]

    Romera, Ad´ equation algorithme architecture pour flot optique sur GPU embarqu´ e, Theses, Sorbonne Universit´ e (2023)

    T. Romera, Ad´ equation algorithme architecture pour flot optique sur GPU embarqu´ e, Theses, Sorbonne Universit´ e (2023)

    work page 2023

  28. [28]

    L. P. N. Mendes, A. M. Ricardo, A. J. M. Bernardino, and R. M. L. Ferreira, A comparative study of optical flow methods for fluid mechanics, Experiments in Fluids 63 (2021)

    work page 2021

  29. [29]

    S. Cai, S. Zhou, C. Xu, and Q. Gao, Dense motion estimation of particle images via a convo- lutional neural network, Experiments in Fluids 60 (2019)

    work page 2019

  30. [30]

    FlowNet: Learning Optical Flow with Convolutional Networks

    P. Fischer, A. Dosovitskiy, E. Ilg, P. H¨ ausser, C. Hazırba¸ s, V. Golkov, P. van der Smagt, D. Cremers, and T. Brox, Flownet: Learning optical flow with convolutional networks (2015), arXiv:1504.06852 [cs.CV]

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  31. [31]

    C. Yu, Y. Fan, X. Bi, Y. fei Kuai, and Y. Chang, Deep dual recurrence optical flow learning for time-resolved particle image velocimetry, Physics of Fluids (2023)

    work page 2023

  32. [32]

    Teed and J

    Z. Teed and J. Deng, Raft: Recurrent all-pairs field transforms for optical flow, in European Conference on Computer Vision (2020)

    work page 2020

  33. [33]

    Hui and C

    T.-W. Hui and C. C. Loy, Liteflownet3: Resolving correspondence ambiguity for more accurate 35 optical flow estimation (2020), arXiv:2007.09319 [cs.CV]

    work page arXiv 2020

  34. [34]

    Lagemann, K

    C. Lagemann, K. Lagemann, S. Mukherjee, and W. Schr¨ oder, Deep recurrent optical flow learning for particle image velocimetry data, Nature Machine Intelligence 3, 641 (2021)

    work page 2021

  35. [35]

    Lagemann, K

    C. Lagemann, K. Lagemann, S. Mukherjee, and W. Schroeder, Generalization of deep recur- rent optical flow estimation for particle-image velocimetry data, Measurement Science and Technology 33 (2022)

    work page 2022

  36. [36]

    Lagemann, K

    C. Lagemann, K. Lagemann, S. Mukherjee, and W. Schr¨ oder, Challenges of deep unsupervised optical flow estimation for particle-image velocimetry data, Experiments in Fluids (2024)

    work page 2024

  37. [37]

    T. C. Redman, If your data is bad, your machine learning tools are useless, Harvard Business Review - Analytics and Data Science (2018)

    work page 2018

  38. [38]

    Gautier and J

    N. Gautier and J. L. Aider, Real-time planar flow velocity measurements using an optical flow algorithm implemented on GPUs, Journal of Visualization 18, 277 (2015)

    work page 2015

  39. [39]

    Gautier and J

    N. Gautier and J. L. Aider, Control of the separated flow downstream of a backward-facing step using visual feedback, Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences 469 (2013)

    work page 2013

  40. [40]

    Gautier and J

    N. Gautier and J. L. Aider, Frequency-lock reactive control of a separated flow enabled by visual sensors, Experiments in Fluids 56, 1 (2015)

    work page 2015

  41. [41]

    Varon, J

    E. Varon, J. L. Aider, Y. Eulalie, S. Edwige, and P. Gilotte, Reactive control of large-scale structures in a turbulent wake (2017)

    work page 2017

  42. [42]

    Giannopoulos and J.-L

    A. Giannopoulos and J.-L. Aider, Data-driven order reduction and velocity field reconstruction using neural networks: The case of a turbulent boundary layer, Physics of Fluids 32 (2020)

    work page 2020

  43. [43]

    Giannopoulos, P.-Y

    A. Giannopoulos, P.-Y. Passaggia, N. Mazellier, and J.-L. Aider, On the optimal window size in optical flow and cross-correlation in particle image velocimetry: application to turbulent flows, Experiments in Fluids 63, 57 (2022)

    work page 2022

  44. [44]

    J. I. Cardesa, A. Vela-Mart´ ın, and J. Jim´ enez, The turbulent cascade in five dimensions, Science 357, 782 (2017)

    work page 2017

  45. [45]

    B. D. Lucas and T. Kanade, An iterative image registration technique with an application to stereo vision, in International Joint Conference on Artificial Intelligence (1981)

    work page 1981

  46. [46]

    A. Giannopoulos, Optical Flow Velocimetry : optimization, benchmarking and application to system identification, modelling and control of shear flows , Theses, Sorbonne Universit´ e (2021). 36

    work page 2021

  47. [47]

    Pimienta Alvernia, High frequency and high resolution Live Optical Flow velocimetry

    J. Pimienta Alvernia, High frequency and high resolution Live Optical Flow velocimetry. Ap- plication to the monitoring and control of fluid flows. , Theses, Universit´ e Paris sciences et lettres (2024)

    work page 2024

  48. [48]

    Mendes, A

    L. Mendes, A. Bernardino, and R. M. Ferreira, PIV-image-generator: An image generating software package for planarPIV and optical flow benchmarking, SoftwareX12, 100537 (2020)

    work page 2020

  49. [49]

    S. Beresh, The influence of velocity gradients on PIV measurements of turbulence statistics: A preliminary study, 26th AIAA Aerodynamic Measurement Technology and Ground Testing Conference (2008), https://arc.aiaa.org/doi/pdf/10.2514/6.2008-3710

    work page doi:10.2514/6.2008-3710 2008

  50. [50]

    Westerweel, On velocity gradients in PIV interrogation, Exp Fluids 44, 831–842 https://doi.org/10.1007/s00348-007-0439-3 (2008)

    J. Westerweel, On velocity gradients in PIV interrogation, Exp Fluids 44, 831–842 https://doi.org/10.1007/s00348-007-0439-3 (2008)

    work page doi:10.1007/s00348-007-0439-3 2008

  51. [51]

    J. J. Gibson, The senses considered as perceptual systems (1967)

    work page 1967

  52. [52]

    J. J. Gibson, The ecological approach to visual perception (1979)

    work page 1979

  53. [53]

    J. J. Gibson, The perception of th visual world (Houghton Mifflin, 1950)

    work page 1950

  54. [54]

    J. J. Gibson, A theory of direct visual perception (2002)

    work page 2002

  55. [55]

    Andalibi, L

    M. Andalibi, L. Hoberock, and H. Mohamadipanah, Effects of texture addition on optical flow performance in images with poor texture, Image and Vision Computing 40, 1 (2015)

    work page 2015

  56. [56]

    Thielicke and R

    W. Thielicke and R. Sonntag, Particle image velocimetry for MATLAB: Accuracy and en- hanced algorithms in PIVlab, Journal of Open Research Software 9, 12 (2021)

    work page 2021

  57. [57]

    J. Saif, M. H. Hammad, and I. A. A. Alqubati, Gradient based image edge detection, Inter- national journal of engineering and technology 8, 153 (2016)

    work page 2016

  58. [58]

    Aguilar-Cabello, L

    J. Aguilar-Cabello, L. Parras, and C. del Pino, Dpivsoft-opencl: A multicore cpu-gpu accel- erated open-source code for 2d particle image velocimetry, SoftwareX 20, 101256 (2022). 37

    work page 2022