On the Use of RBF for Global Field Description in DIC

The standard digital image correlation (DIC) algorithm samples the region of interest (ROI) on a regular grid to describe the displacement field on a specimen. In each sampling point, a local problem is solved. Thus, apart from the oversampling usually used when selecting the grid parameters, the solutions related to each point are statistically independent. This approach is general but usually over-restraining because the measured displacement fields are usually continuous. Using global shape functions may thus give some advantages in terms of accuracy and computational time. The standard way to achieve this objective is using a finite element-like approach, but other options are possible. This work focuses on using radial base functions: various approaches to globally describe the displacement fields are possible, and different families of functions exist. We intend to compare some of them, looking at the performance (accuracy), implementation complexity, and computational time.