Evaluation of crack opening phenomenon using subset-optimized digital image correlation

Abstract

This paper presents crack opening phenomenon evaluation using digital image correlation (DIC) with a statistically optimized subset size. In conventional DIC analysis, the subset sizes varying from several pixels to more than hundred pixels have been often selected by experts' subjective judgement based on conventional subset size determination algorithms. Since these conventional subset size determination algorithms, however, calculate speckle pattern features at a certain location of a single target image, it is difficult to consider not only all speckle pattern features within region of interest (ROI) but also the random measurement noises during the digital image acquisition process. To overcome the technical limitation, a statistical optimization algorithm of the subset size, which calculates the optimal subset size by the 3-loop iteration of normalized cross correlation within the entire ROI, is newly proposed. In addition, the optimal subset-based DIC analysis is applied to crack opening phenomenon evaluation in a mock-up concrete specimen under step loading conditions. The validation test results show 3.6 μm maximum error compared with the ground truth which is obtained by direct measurement, while a conventional subset size determination algorithm-based DIC analysis produces the maximum error of 62.7 μm.