An improved iterative update strategy and high-order shape functions for digital image correlation
Shihai Lan,Tianxiang Ren,Canyu Zhu,Yaru Wang,Yufei Wang,Shangquan Wu,Qingchuan Zhang
DOI: https://doi.org/10.1016/j.optlaseng.2024.108070
IF: 5.666
2024-02-03
Optics and Lasers in Engineering
Abstract:Digital image correlation (DIC) is a widely used technique for measuring displacement fields in various applications. However, traditional DIC algorithms, constrained by the low-order shape function (up to the second order), exhibit certain inherent limitations. The under-match between high-order deformation and low-order shape function causes a non-negligible error in the measurement, which becomes the primary source of the systematic error and limits the measurement accuracy. High-order shape function has not been previously applied due to complex derivation and poor computational efficiency. Consequently, existing algorithms and software cannot realize high-accuracy measurement for complex heterogeneous displacement fields. To address this issue, a high-order DIC algorithm with high computational efficiency is proposed. In this work, a novel iterative update strategy (termed as backward subtractive update strategy) with high efficiency and no requirement for complex derivation is first proposed, serving as the foundation of the high-order DIC algorithm. Numerical simulations demonstrate that the proposed update strategy can attain measurement accuracy comparable to the inverse compositional strategy. Subsequently, the optimization of high-order shape functions is investigated to reduce the deformation parameters, thereby enhancing the computational efficiency. Finally, the backward subtractive Gauss-Newton (BS-GN) DIC method based on the optimized high-order shape functions is proposed for complex deformation measurement. Numerical simulations reveal that the proposed high-order BS-GN DIC method significantly reduces systematic errors, realizing high-resolution measurement for high-frequency complex deformation. Furthermore, the proposed algorithm shows a remarkable improvement in computational efficiency when compared to the high-order FA-GN algorithm.
optics