Assessment of h-refinement procedure for global digital image correlation

Abstract

Digital image correlation (DIC) is a non-interferometric optical technique able to measure bi-dimensional (tri-dimensional in its stereo version) displacement fields by comparing two images acquired before and after motion. In its standard formulation, the measurement is performed by comparing a small subset of pixels in the two images, looking for the best set of parameters that minimize a given error function under the assumption of an affine displacement field. This procedure is performed in several points of the image so as to allow the reconstruction of the full displacement field by interpolating the various independent samplings. Global DIC formulation avoids the a posteriori interpolation by describing the displacement field with a finite-element-like mesh: the resulting minimization is global because the fitting parameters (usually the nodal displacements) are shared by neighbor elements; thus the standard deviation of displacements is lower. However, the new formulation raises the problem of the identification of the optimal mesh. A possible solution to this problem is the adaptive mesh refinement. This work proposes a simple h-refinement approach for global DIC and analyses its performance using both synthetic and experimental data .