2D DIC from single moving camera by homographies compensation

The use of unmanned aerial vehicles in combination with vision-based measurement techniques has increasing interest in structural health monitoring. When displacement fields must be measured, Digital Image Correlation is one of the most popular techniques for image analysis. However, when images are taken from a single camera mounted on a drone, DIC gives incorrect results, since the camera movement creates apparent displacements and deformations. In the paper a technique for the compensation of these perspective distortions is proposed for planar surface case. The technique uses homographies estimated on a limited region of the measurement target assumed undeformed. Tests on numerically generated images have been performed to validate the proposed algorithm and to study the effect of the non-complete compliance with theoretical assumptions. Model for compensation error prediction is also proposed. Finally, some experimental tests considering real targets have been performed, obtaining good performances for the crack aperture case study.