A high-frequency low-cost technique for measuring small-scale water level fluctuations using computer vision

Measuring and monitoring the water depth/level is a key issue when studying free-surface flows. These tasks are usually expensive in terms of time and money; however, even that expenditure is sometimes not enough to assure reliable and/or accurate results. Free-surface flows are complicated to deal with, as the instability caused by e.g., turbulence, wind, or air-entrainment, can cause important spatial and temporal fluctuations at the surface level. This work presents a non-intrusive, computer vision-based image treatment and segmentation technique that assures the detection and measurement of the free-surface water fluctuations along space and time. The laboratory physically based tests under steady flow conditions, and different channel bed roughness and slopes, showed a very good fit with manually direct measurements carried out with a point-gauge micrometre. Under unsteady flow conditions this technique also showed to successfully deal with applications requiring high spatiotemporal resolution of water depth/level measurements.