Phase-augmented digital image correlation for high-accuracy deformation measurement: Theory, validation, and application to constitutive law learning

Abstract

Digital image correlation (DIC) is a prominent technique for full-field, non-contact deformation characterization. Despite its sub-pixel sensitivity for displacement measurement, conventional DIC often suffers from inadequate signal-to-noise ratios (SNRs) when measuring small deformations in stiff and/or brittle materials. This work presents phase-augmented DIC (PA-DIC), a novel method that integrates coherent illumination with image correlation to achieve measurement accuracy that surpasses that of conventional DIC. Unlike conventional DIC, which relies on non-coherent illumination to maintain gray-level conservation during deformation, PA-DIC leverages speckle phase information to measure displacement with improved sensitivity. We applied PA-DIC to characterize rigid-body rotation and non-uniform tensile deformation, validating its high accuracy and reliability. Furthermore, we demonstrated its application for machine learning of an orthotropic elastic constitutive relationship. This was achieved using a hybrid finite element method and neural network (FEM-NN) optimization framework supplied with high-accuracy non-uniform strain data from PA-DIC. With its exceptional measurement accuracy, PA-DIC opens new possibilities for advanced full-field measurement and data-driven material characterization in the small deformation regime.