Pixel-Domain Modal-Parameter-Based Structural Health Monitoring Using In-Building Cameras: A Review

Abstract

Recent studies have shown that existing in-building surveillance cameras can be repurposed as practical sensors for structural health monitoring of buildings. While many vision-based structural monitoring methods aim to recover physical displacements, a smaller but important line of research has shown that useful structural information can also be extracted directly in the pixel domain. This review examines modal-parameter-based structural health monitoring approaches that use cameras installed inside buildings. The reviewed studies are organized into four stages in the development of the topic: proof-of-concept frequency extraction from existing indoor cameras, pixel-based operating-mode identification, synchronized multi-camera modal identification, and post-earthquake damage assessment through frequency-shift analysis. The review highlights the advantages of pixel-domain processing for practical deployment, especially the reduced calibration needs and compatibility with pre-installed cameras, while also discussing key limitations related to synchronization, camera motion, lighting, and the interpretation of pixel-domain outputs. The paper concludes by identifying promising directions for future investigation.