Real-time matching strategy for rotary objects using digital image correlation.

Abstract

Real-time monitoring of structural health conditions for rotary objects is of importance for safety assessments. In this work, an efficient algorithm based on digital image correlation is presented to achieve accurate rotational matching in real time. The proposed algorithm measures rotation in object motion with an integer pixel search followed by a subpixel correlation refinement. In the integer pixel search, the reference subset is rotated inversely to facilitate the correlation computation between the reference and target subsets. Then an independent and global integer pixel search for each point of interest is performed by applying the particle swarm optimization algorithm. Finally, a modified iterative registration algorithm is introduced to refine the displacement in the subpixel level by considering both the rotation angle and displacement components. Simulation and rotation experiments demonstrate that the proposed method achieves rapid and accurate measurements and is an effective method for retrieving the rotation data of rotating structures.