Figure 2. Histogram of an instantaneous PIV image (a) when pixel-locked, and (b) after adjustment.

​Pixel locking in PIV 

Figure 1. Sub-pixel displacement (a) when pixel-locked, and (b) after adjustment.

Project leader : R. Jason Hearst

Main supervisor(s): Prof Ganapathisubramani

Figure 3. Quantification metric of pixel-locking in a particular PIV field. Q = 0 means no pixel-locking. Q = 100% means only integer values are measured.


Pixel-locking is an intrinsic error that occurs in particle image velocimetry (PIV) images as a result of insufficient resolution to accurately determine the sub-pixel particle displacement between images.  When an image is pixel-locked, the sub-pixel displacement is biased towards an integer pixel displacement, see Figure 1(a).  This can be adjusted to a distribution that has no bias towards an integer pixel displacement, Figure 1(b), using a transfer function.  With such an adjustment, some of the peaks in an instantaneous PDF of the velocity distribution in an image are found to be spurious and a result of pixel locking, Figure 2.  

It is also found that pixel-locking is non-uniform across an image, and as a result, adjustment should be made on a vector-by-vector basis rather than en masse, Figure 3.