Key points are not available for this paper at this time.
The predominant method for infrastructure evaluation is visual inspection. The large number and scale of inspections often make it difficult to catalog the location, size, and severity of the identified damage. Several nondestructive methods can be used to detect cracks, such as acoustic emission and ultrasound, but they require the installation of many sensors, typically involve measurements in a predetermined area, and often require post-processing. Though image-based crack detection is limited to inspecting surface cracks, it has advantages including speed, repeatability, and large area coverage. Furthermore, the use of image stitching can provide a comprehensive assessment of the health of the entire structure. This method could also be applied using an unmanned aerial vehicle (UAV). This paper discusses major challenges in automated crack detection and implementation on UAVs.
Ellenberg et al. (Tue,) studied this question.