Methodology and Validation of UAV-Based Video Analysis Approach for Tracking Earthquake-Induced Building Displacements

Abstract

Unmanned aerial vehicle (UAV) imagery has recently emerged as a promising alternative for operational condition inspection and postdisaster damage assessment of civil structures (e.g., bridges and buildings). However, the use of such sensing techniques for quantitatively tracking the subtle (centimeter-level) variations of the responses of structures has been limited. This is largely due to the difficulties related to obtaining accurate location measurements of the cameras aboard small UAV platforms. To address this research gap, we propose a video analysis methodology for tracking the displacement response of buildings subject to dynamic loads using camera-equipped UAV platforms. The movement of the image sensor on-board the UAV platform is corrected to allow image-by-image natural feature detection and tracking. In this methodology, the image processing procedure does not rely on the camera position and orientation. As such, the approach first corrects for image distortion introduced by UAV drift and subsequently extracts the dynamic displacements of the building by tracking its natural features at the pixel level. Motion-tracking errors are investigated by analyzing the building displacements using pre- and postevent videos. The proposed methodology is validated by monitoring the dynamic response of a full-scale building during a shake table test program. Uniquely, global positioning system (GPS) displacement measurements are independently utilized to validate the proposed UAV video-based method and assess its effectiveness for capturing the dynamic responses of full-scale structures with a level of precision that is sufficient for engineering applications (less than 2 cm root-mean-square errors).