Phase Correlation Processing for DPIV Measurements: Part II — Spectral Domain Analysis

A novel Digital Particle Image Velocimetry (DPIV) correlation method is introduced which estimates the displacement using the phase content within the Fourier based cross-correlation. The use of weighted least squares and robust least squares estimation is introduced in order to improve the linear phase estimation of this technique. Spectral filters are constructed using the energy content of PIV images to define weighting functions for the dominant singular vector regressions. This performance of this technique is measured using Monte Carlo simulations of DPIV images. The resulting error analysis demonstrates substantially reduced errors for higher particle-image diameters for images containing low amounts of noise. For high noise images, the reduction in bias and RMS errors is not as drastic due to limitations of extracting the phase information. However, this correlation technique is able to eliminate peak-locking errors, shown to be a substantial source of error in noisy images, by directly extracting the displacement information in the spectral domain.