Subpixel Displacement Matching Accuracy of PIV and the Effects of Particle Size and Concentration

Two subpixel displacement matching methods of PIV, i.e., the quadratic polynomial surface fitting method and subpixel displacement algorithm based on gray gradient, are compared in this paper. In the case where all of the particles move toward the same direction with the same displacement, the subpixel displacements and errors of the two methods are given under different windows, which shows that the displacement accuracy of the quadratic polynomial surface fitting method increases with the increasing of matching windows while the accuracy of subpixel displacement algorithm based on gray gradient shows little variations with the matching windows. The absolute mean displacement errors of the two methods are smaller than 0.1 pixel. For the case of idealized vortex flow, it is found that the displacement accuracy of the quadratic polynomial surface fitting method becomes within 0.2 pixels, and the accuracy of subpixel displacement algorithm based on gray gradient is about 0.05 pixels. Both particle sizes and concentrations influence the matching accuracy, results show that the larger the matching window, the larger the range of the suitable particle concentration, which is believed to be related to the valuable information of the particle image contained in different matching windows. Moreover, the particles with diameter of 3 pixels are found to be suitable for matching windows of 16 pixels×16 pixels, 32 pixels×32 pixels and 64 pixels×64 pixels. The two subpixel matching methods are applied to analyze the actual PIV images obtained in wave tank experiments, both of them can achieve reasonable results.