Application of masked two-dimensional Fourier spectra for improving the accuracy of STIV-based river surface flow velocity measurements

Space-time image velocimetry (STIV) is a well-established image-based technique for measuring river surface flow from videos or image time series. Because the fundamental measurement mechanism is relatively simple, STIV has been used widely in research and for practical purposes in Japan and other countries. The advantage of STIV over other imaging techniques such as large-scale particle image velocimetry (LSPIV) is its robustness. More specifically, the streamwise velocity component can be efficiently estimated by simply measuring the slope appearing in a space-time image (STI) that represents the advection of surface textures. However, in the course of practical applications cases occur in which STIV yields erroneous results when the measurement conditions at the river site are not satisfactory for accurate measurements, such as when the water surface includes textures other than those directly related to turbulent advection. In this research, surface textures are classified into ten types and image filters based on wavenumber–frequency spectra are developed to improve the quality of the STI so that the texture contains only those features that represents the advection of surface turbulence. The quality of the STI can be improved with the filters and thus further increase the robustness of the measurement technique.