Performance characteristics of a wavelet-based optical flow method for velocimetry from tracer particle images

We present a wavelet-based optical flow (WOF-X) method designed for velocimetry from experimental image pairs of tracer particles. The accuracy of the method is assessed and compared to conventional correlation-based PIV using synthetic tracer particle images that are transported by a simulated two-dimensional flow field. The performance of both methods is evaluated for different values of inter-frame particle displacement and particle seeding density. Optimal values for these parameters are found for both methods, and it is observed that the optical flow method out-performs correlation-based PIV in terms of accuracy and spatial resolution; that is, WOF-X produces a dense estimate of the velocity field (i.e. one velocity vector per pixel). The dynamic range is computed for both methods and the sensitivity to out-of-plane displacement of tracer particles also is assessed.