Improving accuracy of image-based depth inversion with an adaptive window optimization
Byunguk Kim, Yong Sung Park, Hyoseob Noh, Minjae Lee · Coastal Engineering Journal · 2025
A depth inversion algorithm is an efficient remote-sensing technique for retrieving nearshore bathymetry information using the propagation of surface waves captured over time in a video. The essence of these algorithms is to retrieve spatial distributions of the wave parameters from wave field videos, and estimate the water depth according to the dispersion relation. While depth inversion algorithms offer the benefits of non-intrusive data collection and cost-effectiveness, they still generally lack the precision of in-situ measurement. To enhance accuracy, this study addresses the impact of the error-inducing factor, namely the interrogation window size, on wave parameter estimation in linear depth inversion. This is accomplished through numerical analysis using a fully nonlinear Boussinesq model with various bathymetry conditions, as well as through field experiment. As a result, an adaptive method for determining window size is proposed considering both the surface wavelength and changes in the bottom elevation. The optimum range of the window-size-to-wavelength ratio according to different bottom slope is suggested, which is validated against the field data.