Digital Stroboscopy Using Event-Driven Imagery

The current structural health monitoring techniques rely on contact-based sensor arrays that are hard to install and expensive to maintain. Video-based dynamics technology for structural health monitoring is being explored in order to optimize cost and improve performance. However, current imagers capture both the dynamic and static portions of a scene when only the dynamic portion is needed. The capture of both static and dynamic portions of a scene leads to unnecessarily large amounts of data. Further innovation has proffered the concept of event-driven imagery and the silicon retina, a neuromorphic imager technology. The silicon retina detects and reports only changes in illumination or light intensity on a pixel by pixel basis, using a time scale that is on the order of tens of microseconds. While the leanness and dynamic capability of event-driven data make it attractive for video-based structural health monitoring, the mathematical framework for performing signal processing and feature extraction from event-driven data is very embryonic. This work focuses on the development of a signal processing framework for generating conventional frames from event-driven imagery that preserve the temporal content of interest. Specifically, this research focused on identifying, preserving, and visualizing the structural dynamic characteristics of structures through the use of virtual shutters and digital stroboscopy. Based on the results of this research, several issues are outlined that can be used as guidelines for an innovative and deployable event imager-based structural dynamics identification system.