Open Access

Daniel B. Bung, Daniel Valero

• Optical flow estimation is known from Computer Vision where it is used to determine obstacle movements through a sequence of images following an assumption of brightness conservation. This paper presents the first study on application of the optical flow method to aerated stepped spillway flows. For this purpose, the flow is captured with a high-speed camera and illuminated with a synchronized LED light source. The flow velocities, obtained using a basic Horn–Schunck method for estimation of the optical flow coupled with an image pyramid multi-resolution approach for image filtering, compare well with data from intrusive conductivity probe measurements. Application of the Horn–Schunck method yields densely populated flow field data sets with velocity information for every pixel. It is found that the image pyramid approach has the most significant effect on the accuracy compared to other image processing techniques. However, the final results show some dependency on the pixel intensity distribution, with better accuracy found for grey values between 100 and 150.