Multi-camera large-scale particle image velocimetry

Abstract

Large-scale particle image velocimetry (LSPIV) has been widely used for quantifying the velocity distribution of environmental water surfaces, especially river flows. In LSPIV, the evaluated velocity distribution of a region far from the camera tends to have a large uncertainty due to strong image distortion and reduced image resolution, especially for directions parallel to the viewing angle. This study proposes a method that overcomes this problem by combining the results of LSPIV measurements of images obtained from different viewing angles. The velocity component for the direction orthogonal to the viewing angle (the 1D velocity component) for each viewing angle was extracted and then the 2D velocity field was reconstructed based on a least squares algorithm. To validate the proposed method, a rigid rotating disk experiment and a circulatory flow flume experiment were conducted. The method was used to reproduce a complex flow formed around an opposite river edge using images obtained from two different locations on a river bank. Circulatory flow was not visible from results obtained from single viewing angles but was successfully represented in the reconstructed velocity field.