Skip to main content
SpringerLink
Log in

Dense motion estimation of particle images via a convolutional neural network

  • Research Article
  • Published:
Experiments in Fluids Aims and scope Submit manuscript

Abstract

In this paper, we propose a supervised learning strategy for the fluid motion estimation problem (i.e., extracting the velocity fields from particle images). The purpose of this work is to design a convolutional neural network (CNN) for estimating dense motion field for particle image velocimetry (PIV), which allows to improve the computational efficiency without reducing the accuracy. First, the network model is developed based on FlowNetS, which is recently proposed for end-to-end optical flow estimation in the computer vision community. The input of the network is a particle image pair and the output is a velocity field with displacement vectors at every pixel. Second, a synthetic dataset of fluid flow images is generated to train the CNN model. To our knowledge, this is the first time a CNN has been used as a global motion estimator for particle image velocimetry. Experimental evaluations indicate that the trained CNN model can provide satisfactory results in both artificial and laboratory PIV images. The proposed estimator is also applied to the experiment of turbulent boundary layer. In addition, the computational efficiency of the CNN estimator is much superior to those of the traditional cross-correction and optical flow methods.

Graphical abstract

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2

[This figure is modified from the paper Dosovitskiy et al. (2015)]

Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13

Similar content being viewed by others

Notes

  1. The original FlowNetS receives RGB images as input, hence six channels in all are required. However, PIV images are typically created using monochrome cameras. Therefore, two-channel input for this CNN is also sufficient.

  2. Available online: https://github.com/mrahrauld/PIVLab.

References

  • Adrian R, Westerweed J (2011) Particle image velocimetry. Cambridge University Press, Cambridge

    Google Scholar 

  • Brox T, Bruhn A, Papenberg N, Weickert J (2004) High accuracy optical flow estimation based on a theory for warping. In: Proceedings of the European conference on computer vision, pp 25–36

    Google Scholar 

  • Bruhn A, Weickert J, Schnörr C (2005) Lucas/Kanade meets Horn/Schunck: combining local and global optic flow methods. Int J Comput Vis 61(3):211–231

    Article  Google Scholar 

  • Cai S, Mémin E, Dérian P, Xu C (2018) Motion estimation under location uncertainty for turbulent fluid flows. Exp Fluids 59(1):8

    Article  Google Scholar 

  • Carlier J (2005) Second set of fluid mechanics image sequences. European project fluid image analysis and description (FLUID). http://www.fluid.irisa.fr

  • Champagnat F, Plyer A, Le Besnerais G, Leclaire B, Davoust S, Le Sant Y (2011) Fast and accurate PIV computation using highly parallel iterative correlation maximization. Exp Fluids 50(4):1169–1182

    Article  Google Scholar 

  • Chen X, Zillé P, Shao L, Corpetti T (2015) Optical flow for incompressible turbulence motion estimation. Exp Fluids 56(1):1–14

    Article  Google Scholar 

  • Chen L, Papandreou G, Kokkinos I, Murphy K, Yuille A (2018) Deeplab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. IEEE Trans Pattern Anal Mach Intell 40(4):834–848

    Article  Google Scholar 

  • Corpetti T, Mémin E, Pérez P (2002) Dense estimation of fluid flows. IEEE Trans Pattern Anal Mach Intell 24(3):365–380

    Article  Google Scholar 

  • Corpetti T, Heitz D, Arroyo G, Mémin E, Santa-Cruz A (2006) Fluid experimental flow estimation based on an optical-flow scheme. Exp Fluids 40(1):80–97

    Article  Google Scholar 

  • Dao X, Collewet C (2012) Drag reduction of the plane poiseuille flow by partitioned visual servo control. In: Proceedings of the American control conference, pp 4084–4089

  • Dosovitskiy A, Fischer P, Ilg E, Hausser P, Hazirbas C, Golkov V, Van der Smagt P, Cremers D, Brox T (2015) Flownet: learning optical flow with convolutional networks. In: Proceedings of the IEEE international conference on computer vision, pp 2758–2766

  • Heitz D, Mémin E, Schnörr C (2010) Variational fluid flow measurements from image sequences: synopsis and perspectives. Exp Fluids 48(3):369–393

    Article  Google Scholar 

  • Horn B, Schunck B (1981) Determining optical flow. Artif Intell 17(1–3):185–203

    Article  Google Scholar 

  • Ilg E, Mayer N, Saikia T, Keuper M, Dosovitskiy A, Brox T (2017) Flownet 2.0: evolution of optical flow estimation with deep networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, vol 2

  • Kingma D, Ba J (2014) Adam: a method for stochastic optimization. In Proceedings of the 3rd international conference on learning representations (ICLR)

  • Krizhevsky A, Sutskever I, Hinton G (2012) Imagenet classification with deep convolutional neural networks. Adv Neural Inf Process Syst 25(2):1097–1105

    Google Scholar 

  • Lee Y, Yang H, Yin Z (2017) PIV-DCNN: cascaded deep convolutional neural networks for particle image velocimetry. Exp Fluids 58(12):171

    Article  Google Scholar 

  • Li Y, Perlman E, Wan M, Yang Y, Meneveau C, Burns R, Chen S, Szalay A, Eyink G (2008) A public turbulence database cluster and applications to study Lagrangian evolution of velocity increments in turbulence. J Turbul 9(31):1–29. https://doi.org/10.1080/14685240802376389

    Article  MATH  Google Scholar 

  • Liu T, Shen L (2008) Fluid flow and optical flow. J Fluid Mech 614:253–291

    Article  MathSciNet  Google Scholar 

  • Liu T, Merat A, Makhmalbaf H, Fajardo C, Merati P (2015) Comparison between optical flow and cross-correlation methods for extraction of velocity fields from particle images. Exp Fluids 56(8):166

    Article  Google Scholar 

  • Long J, Shelhamer E, Darrell T (2015) Fully convolutional networks for semantic segmentation. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 3431–3440

  • Mayer N, Ilg E, Fischer P, Hazirbas C, Cremers D, Dosovitskiy A, Brox T (2018) What makes good synthetic training data for learning disparity and optical flow estimation? Int J Comput Vis 1:1–19

    Google Scholar 

  • Perlman E, Burns R, Li Y, Meneveau C (2007) Data exploration of turbulence simulations using a database cluster. In Proceedings of the 2007 ACM/IEEE conference on supercomputing

  • Qian N (1999) On the momentum term in gradient descent learning algorithms. Neural Netw 12(1):145–151

    Article  MathSciNet  Google Scholar 

  • Raffel M, Willert C, Wereley S, Kompenhans J (2007) Particle image velocimetry: a practical guide. Springer, Berlin

    Google Scholar 

  • Resseguier V, Mémin E, Chapron B (2017) Geophysical flows under location uncertainty, part II: quasigeostrophic models and efficient ensemble spreading. Geophys Astrophys Fluid Dyn 111(3):177–208

    Article  MathSciNet  Google Scholar 

  • Scarano F (2002) Iterative image deformation methods in PIV. Meas Sci Technol 13(13):R1

    Article  Google Scholar 

  • Stanislas M, Okamoto K, Kähler CJ, Westerweel J (2005) Main results of the second international PIV challenge. Exp Fluids 39(2):170–191

    Article  Google Scholar 

  • Stanislas M, Okamoto K, Kähler CJ, Westerweel J, Scarano F (2008) Main results of the third international PIV challenge. Exp Fluids 45(1):27–71

    Article  Google Scholar 

  • Sun D, Roth S, Black M (2010) Secrets of optical flow estimation and their principles. In: Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR), pp 2432–2439

  • Sun Y, Wang X, Tang X (2013) Deep convolutional network cascade for facial point detection. In: Proceedings of IEEE conference on computer vision and pattern recognition, pp 3476–3483

  • Tieleman T, Hinton G (2012) Lecture 6.5-rmsprop: divide the gradient by a running average of its recent magnitude. COURSERA: Neural Netw Mach Learn 4(2):26–31

    Google Scholar 

  • Wereley S, Gui L, Meinhart C (2002) Advanced algorithms for microscale particle image velocimetry. AIAA J 40(40):1047–1055

    Article  Google Scholar 

  • Xu C, Schuster E, Vazquez R, Krstic M (2008) Stabilization of linearized 2D magnetohydrodynamic channel flow by backstepping boundary control. Syst Control Lett 57(10):805–812

    Article  MathSciNet  Google Scholar 

Download references

Acknowledgements

This work was supported in part by the National Natural Science Foundation of China under Grant 61473253, the Foundation for Innovative Research Groups of the National Natural Science Foundation of China under Grant 61621002 and the Fundamental Research Funds for the Central Universities.

Author information

Authors and Affiliations

  1. State Key Laboratory of Industrial Control Technology and the Institute of Cyber-Systems and Control, Zhejiang University, Hangzhou, 310027, Zhejiang, China

    Shengze Cai & Chao Xu

  2. School of Mathematical Sciences, Zhejiang University, Hangzhou, 310027, Zhejiang, China

    Shichao Zhou

  3. School of Aeronautics and Astronautics, Zhejiang University, Hangzhou, 310027, Zhejiang, China

    Qi Gao

Authors
  1. Shengze Cai
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Shichao Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Chao Xu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Qi Gao
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Chao Xu.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

A data sets of particle images for training

A data sets of particle images for training

As introduced in Sect. 3.2, in order to train the neural network, we generate a PIV dataset with more than 13 thousand items (where each item consists of an image pair and a ground truth motion field). The dataset includes various flow motion structures generated by computational fluid dynamics (CFD). Specifically, cases such as Uniform, Back-step, Cylinder are simulated by our own. DNS turbulence and surface quasi-geostrophic (SQG) cases are provided in Carlier (2005) and Resseguier et al. (2017), respectively. Finally, the Johns Hopkins Turbulence Databases can be found in Li et al. (2008), Perlman et al. (2007). Details of the dataset are given in Table 4. The data as well as the trained PIV-NetS model can be requested from the authors and will be accessed online soon.

Table 4 Data sets of particle images for neural network training
Full size table

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cai, S., Zhou, S., Xu, C. et al. Dense motion estimation of particle images via a convolutional neural network. Exp Fluids 60, 73 (2019). https://doi.org/10.1007/s00348-019-2717-2

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s00348-019-2717-2

Search

Navigation