Digital Image Correlation with Dynamic Subset Selection

Abstract

The quality of the surface pattern and selection of subset size play a critical role in achieving high accuracy in Digital Image Correlation (DIC). The subset size in DIC is normally selected by testing different subset sizes across the entire image, which is a laborious procedure. This also leads to the problem that the worst region of the surface pattern influences the performance of DIC across the entire image. In order to avoid these limitations, a Dynamic Subset Selection (DSS) algorithm is proposed in this paper to optimize the subset size for each point in an image before optimizing the correlation parameters. The proposed DSS algorithm uses the local pattern around the point of interest to calculate a parameter called the Intensity Variation Ratio (Λ), which is used to optimize the subset size. The performance of the DSS algorithm is analyzed using numerically generated images and is compared with the results of traditional DIC. Images obtained from laboratory experiments are also used to demonstrate the utility of the DSS algorithm. Results illustrate that the DSS algorithm provides a better alternative to subset size “guessing” and finds an appropriate subset size for each point of interest according to the local pattern.