Skip to Main Content
Previous experiences during earthquake events emphasize the need for new technologies for real-time monitoring and assessment of facilities with high-value nonstructural contents. Moreover, there is a substantial limitation in our ability to rapidly evaluate and identify potential hazard zones within a structure, exposing rescue workers, society, and the environment to unnecessary risks. A real-time image-based monitoring system, which is integrated with warning systems, would allow for improved channeling of resources and informed decision making for rescue workers and building owners. In recognition of these issues, in this paper, we describe a methodology for image-based tracking of seismically induced motions. The methodology includes the acquisition, calibration, and processing of image sequences to detect and track object features under seismic-event conditions. We address the issue of providing a reliable feature/object-detection and object-tracking methodology for an image sequence from a single camera view. In addition, we introduce an extension to the 2-D tracking approach by providing a 3-D feature tracking methodology when the camera array itself is affected by the seismic event. The methods presented are demonstrated using the data collected during the full-scale field vibration tests conducted on a vacant building that was damaged during the 1994 Northridge earthquake (presented in a companion paper). We present experimental tracking results of the implemented algorithms for a variety of objects and discuss additional challenges that emerge when image-based systems are used under these extreme conditions.