Department of Mathematics and Computer Science

Dr. Gilles Labonté, BSc, MSc, PhD Professor

Telephone: 613- 541-6000 ext 6093 

Fax: 613-384-5792 

E-mail: labonte-g@rmc.ca 

Office: Girouard 320 

Particle Image Velocimetry

The Principle of PTV

Particle Tracking Velocimetry (PTV) is one of the most important techniques used to determine the velocity field inside a moving fluid.  A typical PIV experiment is as follows.

• A cloud of very small particles is injected in the fluid.
• A laser beam illuminates a thin plane region of the fluid. 

• 2 photos of the illuminated particles are taken a short time interval Dt apart (a closely related technique consists in using a pulsed laser to produce multiple "instantaneous" exposures on a same photo, but this entails the additionnal problem of having to determine the exposure to which the images belong). 

Photo No 1 taken at t1

Photo No 2 taken at t1 + Dt

The displacements of the particles during Dt can be measured if the images corresponding to the same particles can be recognized in the two pictures, that is, if the correspondence problem of PTV can be solved. These displacements are thereafter used to calculate the average velocities of the particles according to the formula  D/Dt where D is the displacement of the particle.

Solving the correspondence problem however is not trivial because the particle images are all identical.

Further complication: Some particles leave the light sheet during Dt and some others enter it - - - - so that some images appear only in one of the two photos.

Data Processing Stages of PIV

There are three main data processing stages in a PIV experiment: 

• Preprocessing of photographs in order to isolate the images of the particles from the background and the calculation of their center of intensity.  
• Solution of the correspondence problem described above and calculation of the displacements of the particles.  Experiments are divided in two types.
  • Those in which the density of particles is low enough that it is possible to establish the correspondence between individual particles (this is Particle Tracking Velocimetry).
  • Those where the density of particles is larger and in which the correspondence is sought between whole patches of the two photographs, usually by some form of cross-correlation.
• Postprocessing of the measured displacements, in order to reject those that are inconsistent with the known continuity properties of the velocity field.

Some Relevant Internet Sites

Some of the links below lead to a site belonging to an entity not subject to the Official Languages Act. Information on this site is available in the language of the site.

The following webpages contain fundamental information on Particle Image Velocimetry.  Standard images, for testing PIV algorithms, can be downloaded at these Internet sites.  

• PivNet is a network of research centers, universities and industries, financed by the European Community  
• The Center for Quantitative Visualization / Graduate Aeronautical Laboratories of CALTECH  
• At the University of Tokyo  
• The Laboratory for Turbulence and Complex Flow (LTCF) / Department of Theoretical and Applied Mechanics at the University of Illinois at Urbana-Champaign.

My Articles on PIV

• Labonté G. (1998) "A SOM Neural Network That Reveals Continuous Displacement Fields" In the Proceedings of the 1998 International Joint Conference on Neural Networks at WCCI'98 (the World Congress on Computational Intelligence), held at Anchorage Alaska, published by IEEE, ISBN 0-7803-4862-1.  
• Labonté G.(1999): "A New Neural Network for Particle-Tracking Velocimetry" Experiments in Fluids, 26 (4): 340-346.  
• Labonté G. (2000): "On a Neural Network that Performs an Enhanced Nearest-Neighbor Matching", to appear in Pattern Analysis and Applications.