Displacement estimation using a modified seed prediction algorithm in digital image correlation

A new model based strategy for predicting the initial guess in digital image correlation (DIC) analysis is presented. This is accomplished by integrating the physics of deformation with measurements from images using the Bayesian framework. For the purpose of illustration synthetic and experimental data from problems of rigid body motion and finite strains are considered. The proposed algorithm is validated by comparing the predicted deformations with those obtained using the open-source software, Ncorr. The method is successful in accurately capturing the displacements and also achieved a 50% reduction in the computation time in contrast to the existing correlation based approaches. It is anticipated that the formulation presented in this paper will form the basis for developing efficient modelling based computational strategies for DIC applications.