Examination of Optimal Subset Size Selection in Digital Image Correlation

Abstract

This study proposes a method for determining an optimal subset size for digital image correlation. The optimal subset size is defined as that minimizes measurement error. Estimation equation of optimal subset size is proposed based on existing theoretical error model. The estimation equation is a function of 2nd-order displacement gradient (state of deformation), standard deviation of image noise (conservation of intensity values before and after deformation), and mean intensity gradient (speckle pattern quality) as parameters. Measurement errors for various subset sizes are computed in numerical simulations using synthetic speckle images with various measurement condition. For each parameter change, the optimal subset size derived from the measurement error and its estimation equation varies similarly. The analysis of computer-generated speckle images imposed FEM-displacement simulating tensile tests are also conducted and the results indicate the validity of the estimation equation.