Event-Based Imaging Velocimetry - An Assessment Of Event-Based Cameras For The Measurement Of Fluid Flows

Abstract

Event-based vision is a new upcoming field within the field of computer vision. Unlike conventional frame-based imaging, event-based vision (or dynamic vision sensing) asynchronously records binary signals of contrast changes at the pixel level with microsecond resolution. Event-based imaging of small particles illuminated by a laser or other continuous light source generates streaks of events in a space-time domain. From this data the particles' position in both space and time can be inferred. The present work explores the possibilities of harnessing the potentials of event-based vision for fluid flow measurement with focus on two-dimensional, two-component velocity field estimationat particle image densities corresponding to that of conventional PIV. Three different motion detection schemes are implemented. Measurements of a cylinder wake flow demonstrate that EBIV provides time resolved velocity data that is suitable for the retrieval of flow statistics as well as spectral and modal analysis. The present work is an extension of the recently published Exp. Fluids article by the authors on event-based imaging velocimetry (Willert & Klinner, 2022, DOI 10.1007/s00348-022-03441-6) and provides further details on current limitations related to event-based imaging.