Monitoring global earthquake-induced demands using vision-based sensors

A vision-based approach is evaluated for its applicability as a new sensing technology for measuring earthquake-induced motions. The approach evaluated in this paper is advantageous since it requires very limited physical attachment to the structure of interest, is high-speed, high-resolution, and does not introduce additional mass or otherwise modify the properties of the structure. A demonstration experiment is described in which four digital high-speed, high-resolution, chargecoupled-device cameras outfitted with red light-emitting diodes are used to track 21 reflective (nearly) mass less spherical elements discretely mounted on a scale five-story steel frame structure. The structure is mounted on a large bi-axial shake table and subjected to different earthquake motions. A total of eleven conventional (wired) transducers [linear variable displacement transducers and accelerometers] are also discretely mounted on the structure, providing a unique comparison with the vision-based approach. Results from these experiments show that the nonintrusive vision-based approach is extremely promising in terms of its ability to capture inter-story drift, floor level velocities, and accelerations, provided proper post-processing of the dynamic data occurs.