Sparse Observation Assimilated Digital Image Correlation for High Fidelity Deformation Field Reconstruction

Digital image correlation (DIC) is an image-based deformation measurement method, which has been widely used in multiple cutting-edge fields, benefitted from its noncontact, low cost and high robustness. However, achieving higher accuracy and overcoming the tradeoff between resolution and measurement uncertainty have been recognized as bottleneck problems for DIC. This paper proposes a new method by assimilating sparse observations into DIC, constituting “Assimilated DIC”, to reconstruct deformation fields with high fidelity. In the proposed method, the subpixel registration process in the traditional DIC procedure is viewed as a dynamic model, which constitutes the basis for assimilation. Meanwhile, the sparse observation provides highly accurate guidance for correcting the dynamic model, thus facilitating the update of model data towards the ground-truth. In addition to effectively improving the measurement accuracy without lowering the computational efficiency, Assimilated DIC can obtain a higher resolution without compromising the accuracy performance. And the boundary regions of multi-connected structures, which are generally subject to large errors by traditional methods, can be precisely reconstructed by Assimilated DIC. In this work, the proposed method with two types of observations on strain and displacement has been experimentally validated under various deformation cases, including a rigid-body translation, a heterogeneous deformation, and an extension of a holed plate. These excellent properties of Assimilated DIC have been confirmed in comparison with those of a traditional DIC method.