Mesh-based quadrature optimization for accurate and efficient FE-DIC on unstructured triangular meshes

In finite element-based digital image correlation (FE-DIC) with unstructured triangular meshes, accurate evaluation of image-based integrals is critical to robustness and measurement accuracy. Two evaluation point generation strategies are investigated herein, namely, a pixel-based approach and a mesh-based approach. A virtual experiment with a prescribed displacement field and a double-cleavage-drilled compression (DCDC) test on plaster are selected as benchmarks to assess both strategies. The mesh-based approach yields kinematic fields and uncertainties in close agreement with those of the pixel-based approach, while providing measurable computational savings through controlled undersampling of the density parameter. Additionally, damage-informed regularization mitigates the ill-posedness of the DIC problem, and remains close to the reference solution under substantial undersampling. These results show that the mesh-based approach is an efficient alternative for FE-DIC on unstructured meshes when the sampling density is carefully selected for the mesh under consideration.