Author(s): Guerrero Massimo; Lamberti Alberto
Linked Author(s): Massimo Guerrero, Alberto Lamberti
Keywords: Mean velocity field; Digital images; Pattern recognition; Flow streamfunction 1
Abstract: Aim of the work is to acquire the mean depth velocity field during hydrodynamic physical simulation from image processing of dye cloud tracks. The developed tool is a cheap way to automatically acquire the planar permanent mean velocity field of the entire physical model. Five separate steps form the procedure: 1° digital images of dye clouds are acquired with a commercial digital photo camera, 2° homographic relations are used in order to correct images distortion, 3° patterns of dye clouds are recognized inside images frame at consequent fixed time steps, 4° velocities of cloud contours are assessed by comparing consequent patterns, 5 ° flow velocities are assessed with a numerical interpolation of the two-dimensional conservative flow streamfunction. To improve results and decrease computational time, the method needs some tricks during images acquisition and in the following calibration of the Matlab_R_ program. Recognition of the dye clouds depends mainly on colours contrast between dye tracks and model basin. In order to decrease computational effort the time step between images and the grid mesh can be increased but results are sensitive to both parameters, for this reason a compromise between computational time and spatial resolution of velocity acquisition must be achieved caring out a sensitive analysis of spatial and time steps on resulting velocity map.
Year: 2007