A highly accurate iterative PIV technique using a gradient method

An iterative particle image velocimetry (PIV) technique which combines the iterative cross correlation technique and three-point Gaussian peak fitting for a sub-pixel analysis can improve the spatial resolution and accuracy of measurement. The root mean square (RMS) error of the technique is only 0.04 pixels. However, a large interrogation window, typically 32 x 32 pixels or larger, must be taken in order to achieve the high sub-pixel accuracy. This is not compatible with high spatial resolution. In this paper, a new high-resolution PIV technique based on a gradient method is proposed. Initially, the pixel unit displacement is detected by the iterative method. Then, the sub-pixel displacement is evaluated by the use of the gradient method instead of three-point Gaussian peak fitting. The error of the proposed technique has been analytically assessed by Monte Carlo simulations. The RMS error is of the order of 0.01 pixels even with a small interrogation window, 13 x 13 pixels or less. Thus, the method achieves high sub-pixel accuracy and high spatial resolution compatibility. (Original abstract)