Phase Correlation Processing for DPIV Measurements: Part I — Spatial Domain Analysis

Abstract

A novel Digital Particle Image Velocimetry (DPIV) correlation method is presented using nonlinear filtering techniques coupled with a phase-only filter (POF) generalized cross-correlation (GCC). Error analysis demonstrates the deficiency of the GCC alone to successfully improve measurement accuracy, as shown from a decrease in valid vector detection as well as an increase in bias and RMS errors. The use of spatial filters is commonly argued to be disadvantageous for standard Fourier based cross-correlation analysis [1]. However, here we demonstrate that appropriate spatial filters applied to the GCC provide vastly superior performance. The zeropadding method as well as a new spatial filter derived from the Blackman window is introduced, both of which are able to enhance the vector detection capabilities of the GCC processor. In addition, a Gaussian transform filtering technique is introduced which dramatically increases the subpixel resolution of the GCC processor, alleviating apparent effects of peak-locking. The use of these filters with the GCC is able to provide striking improvements in DPIV image correlation.