Pointwise Digital Image Correlation Using the Genetic Algorithm Optimization Method

In this paper, the use of the Genetic Algorithm (GA) optimization method for measuring displacement fields with the Digital Image Correlation (DIC) technique is investigated. The conventional kinematic description of deformations within a subset are restricted to constant displacements along with constant first or second order displacement gradients throughout the subset, which effectively acts as a smoothing filter on higher order displacement fields. Using the GA approach, displacements can be described independently, and therefore discontinuously, for each pixel (i.e., pointwise) within the subset. The resulting technique is called Pointwise Digital Image Correlation, and it eliminates the smoothing attributed to the conventional kinematic description. The effects of varying the parameters for the GA optimization method on the optimized correlation function value are determined for this new technique. Bilinear interpolation is used to obtain subpixel resolution of the displacement fields. The effects of different constraint conditions on the admissible displacements for adjacent pixels are also studied. It is shown that an optimum correlation function value, an order of magnitude smaller than obtained with the conventional DIC technique, can be achieved by choosing a proper combination of GA parameters and constraint conditions. However, there are random variations in the associated displacement fields that may be eliminated with a smoothing filter in order to obtain physically meaningful results when measuring discrete material or structural phenomena that produce rapidly varying and/or discontinuous deformations.