Large deformation measurement using digital image correlation: a fully automated approach.

In digital image correlation, the iterative spatial domain cross-correlation algorithm is considered as a gold standard for matching the corresponding points in two images, but requires an accurate initial guess of the deformation parameters to converge correctly and rapidly. In this work, we present a fully automated method to accurately initialize all points of interest for the deformed images in the presence of large rotation and/or heterogeneous deformation. First, a robust computer vision technique is adopted to match feature points detected in reference and deformed images. The deformation parameters of the seed point are initialized from the affine transform, which is fitted to the matched feature points around it. Subsequently, the refined parameters are automatically transferred to adjacent points using a modified quality-guided initial guess propagation scheme. The proposed method not only ensures a rapid and correct convergence of the nonlinear optimization algorithm by providing a complete and accurate initial guess of deformation for each measurement point, but also effectively deals with deformed images with relatively large rotation and/or heterogeneous deformation. Tests on both simulated speckle images and real-world foam compression experiment verify the effectiveness and robustness of the proposed method.