A PGD approach to reduce the dimensionality of FE-based digital image correlation
Résumé
A new global DIC algorithm based on the proper generalized decomposition is proposed to compute the displacement field between two digital images. The basic idea is to seek the unknown displacement as a sum of products of separated dimensions functions. The method leads to solving one-dimensional problems (along with unidimmensional finite element meshes) in each dimension separately. At each iteration of the correlation problem, a new best rank-one approximations is computed to enhance the displacement approximation. Many examples will be proposed to examplify the performance of the method. Eventually, the case of non rectangular domain is studied. A simple correction is proposed to enable the proposed method to handle arbitrary roi shapes.