Digital image correlation through image registration in the frequency domain

Abstract

The present work analyses displacement field computations for several digital image correlation test cases obtained through the application of image registration in the Fourier domain. The applied method relies on the properties of the discrete Fourier transform with regard to the cross-correlation function to find the rigid displacement between two subsets. The computed displacement fields were compared to their analogous values obtained through a commercial digital image correlation code. The test cases cover both two-dimensional and three-dimensional digital image correlation and make use of the ability of the developed method to calculate the translation and rotation between subsets from images at different loading states, as well as its capacity to match subsets from two cameras of a stereo rig. The presented analysis aims to contribute to the investigation of the use of correlation in the frequency domain as the basis for an alternative to conventional digital image correlation which could provide advantages in terms of robustness and efficiency, particularly in field measurements, where the lack of control over lighting and other factors impacting image quality requires a system which is optimised towards low dependence on the optimal quality of these conditions.