An improvement in the nine-point central difference image correction method for digital particle image velocimetry recording evaluation

A proper adjustment of the translation portion of pixel shifts for the multi-pass, nine-point (9P), central difference image correction (CDIC) method is found to be effective for further reducing the evaluation errors of digital particle image velocimetry recordings. In the improved central difference image correction scheme, the translation portion of pixel shifts is adjusted between the mean value of particle image displacements (initially introduced for the 9P method) and the particle image displacement at the centre (initially suggested for the four-point method) of the interrogation window, so that the evaluation errors of the iterated evaluation may converge to a lower level than both the initially introduced nine- and four-point methods. Based on a 50% overlap of interrogation windows, five possible algorithms to determine the pixel translation shift of the central difference image correction are discussed. Tests with synthetic particle image velocimetry recording pairs of simulated periodic flows demonstrate that the evaluation error of the 9P-central difference image correction method can be reduced to less than half of its initially introduced level, and it is obviously lower than that of the four-point method suggested by Wereley and Gui (2001 4th Int. Symp. on Particle Image Velocimetry (Gottingen, Germany, 17–19 Sept.); 2003 Exp. Fluids 34 42–51). Simulations and real image tests indicate that the error distribution and the peak-locking effect of the central difference image correction method have a two-pixel period.