Background <p>Subset-based image/stereo matching has been widely used in digital image correlation (DIC) for surface shape and deformation measurement. However, because of the inherent inability of regular shape functions to model discontinuous deformation, this standard DIC algorithm cannot provide accurate measurements or even fails around discontinuities, including discontinuous geometries (e.g., boundaries, complex geometries) and discontinuous deformations (e.g., cracks, shear bands).</p> Objective <p>To address this challenge, an easy-to-implement yet practical adaptive subset-subdivision approach is proposed to realize accurate DIC measurement on discontinuous shape and deformation.</p> Methods <p>The method first performs regular DIC calculations and identifies calculation points around discontinuities based on a predefined correlation coefficient threshold. With the assumption that a small part of the subsets near discontinuities can be depicted by regular shape functions, these discontinuity points can be recalculated through an adaptive subset-subdivision strategy. The proposed method eliminates the need to accurately define the discontinuity line, enabling full-automatic, efficient and accurate shape and deformation measurement across geometry and deformation discontinuities.</p> Results <p>Experimental results on both simulated and real-world datasets, including crack propagation, structural deformation, and complex shape reconstruction, demonstrated the efficacy and practicality of the proposed adaptive subset subdivision-based DIC approach.</p> Conclusion <p>The proposed adaptive subset-subdivision strategy offers an easy-to-implement extension of standard DIC to accurately measure discontinuous shape and deformation, and can be implemented as an upgrade within existing DIC frameworks.</p>