A robust motion estimation algorithm for PIV

Abstract

The present study proposes an approach for robust estimation of the instantaneous motion field from a time-lagged pair of PIV images. The method is based on phase correlation, where the phase of the Fourier components is used for motion parameter estimation. Unlike the cross correlation-based techniques, this technique uses ‘whitening’ FIR filters to sharpen the cross correlation maxima, thereby improving the accuracy of identification of the peak. The proposed method also combines the advantages of the phase correlation and the cross correlation techniques in determining the reliability of the estimates, thus providing a method of filtering out a significantly large number of spurious vectors. This reliability metric helps reduce the possibility of over-smoothing the flow field when performing data validation. With regard to the efficiency of the technique, both phase correlation and cross correlation are derived from the Fourier components of the same image region. Each of the estimates can thus be obtained in parallel, without increasing the computational complexity of the system. Unlike many region-based methods that are currently available, the entire motion is decomposed as a global and a local motion field, which helps accurately obtain high interrogation resolution estimates for the local motion field.