Application of the Optical Flow Method to Velocity Determination in Hydraulic Structure Models

Optical flow estimation is used in Computer Vision for detection of moving obstacles in a sequence of images. The optical flow (OF) is defined as the displacement of brightness patterns between two sequent images. In this paper, this method is applied to high-speed images taken in hydraulic structure models for determination of velocity fields. Water is seeded with tracer particles in two cases and is self-aerated in a third case. It will be shown that the OF method gives valuable results which compare well to other velocity measurements, e.g. by Particle Image Velocimetry. The OF method is generally more time-consuming compared to PIV, but an advantage is given by the density of information when using a so-called global method as velocity information is obtained at every pixel location.