A neural net approach in analyzing photograph in PIV

In particle image velocimetry (PIV), photographs of images of the particles in a fluid flow are taken a short interval apart and the velocity field is then determined by measuring the distance that individual particles moved during that time. Attributes of the particles were collected and fed to a neural net to match the particles in the photographs so that the velocity can be measured. The authors consider images with a low concentration of particles so that the discrete images of particles appear as opposed to speckle patterns. It is assumed that the motion of the individual particles is completely random and the velocity of each individual particle is to be found. Results obtained are good for images with particles fairly well spread out