In order to process multi-exposure datasets a dense image autocorrelation approach can be used. This approach computes autocorrelation coefficients for a small patch around every pixel of the input image. Based on location of the peaks in the correlated images one can determine displacements of substructures.
Figure: Top Left: Flat-field corrected input image depicting repetitive spray droplets. Top Right: Correlation peaks analysis. The peak corresponding to the best displacement match is shown as a red marker. Bottom Left: Correlation coefficients for each pixel. Bottom Right: Displacement amplitude computed from the positions of correlation peaks.