next up previous contents
Next: Pattern Matching - the Up: MatPIV behind the scenes Previous: MatPIV behind the scenes   Contents

How does it work?

The fundamentals of Particle Image Velocimetry relies on basic pattern matching. This is a topic touched upon by most introductory books on image processing (see for example Gonzales and Woods, 1992). The important thing to understand about PIV is that it is nothing but an application of these techniques to experiments in fluid (or solid) mechanics. It is very easy to get confused over the fact that to use PIV you need a grasp of digital cameras, particles, lasers (or other illumination devices) as well as the physics of your experiment. In the present text we aim to emphasize that you should treat the PIV technique as a generic tool to measure your experiments. It is generic in the way that we apply principles of pattern matching (which is a software-issue) and these principles have nothing to do with the way we do our experiment (which depends on hardware-issues). The first step is to add some tracer into our flow field. To measure the motion of this tracer we normally have to illuminate it and film it with a camera. Since most cameras record on a 2-dimensional plane, we usually restrict our illumination to a 2-dimensional plane as well - hereafter known as a ``light sheet''. To accomplish this we may need to use optics, and to film it we may need some technological insight (like knowing how to use a video camera), but at the end of the day we go through all these steps in order to generate a pattern in our flow field and registering the motion of it. This pattern is subsequently used for pattern matching which will tell us something about displacements. If we then divide the displacement by the time separation between our images (we assume we have taken two images in order to match patterns between them) we end up with a velocity.



Subsections
next up previous contents
Next: Pattern Matching - the Up: MatPIV behind the scenes Previous: MatPIV behind the scenes   Contents
Johan K. Sveen 2004-08-06