Applications of Video and Signal Processing in Monitoring Transportation Infrastructure Health

Abstract: This study explores the application of advanced video analysis and signal processing techniques in assessing the health of transportation infrastructure. By leveraging a vision-based sensor methodology, video and computer vision algorithms detect and track displacements and magnify small motions across various structural assets, including pedestrian bridges, light poles, traffic signals, and building roofs. These techniques allow us to monitor critical structural elements that may otherwise go unnoticed but are essential for maintaining safety and efficiency within transportation networks.
A key focus is detecting early signs of potential structural deficiencies by analyzing modal parameters such as natural frequencies and damping ratios. Natural frequencies, which relate to a structure's stiffness and mass distribution, often signal issues like stiffness degradation or damage when they shift over time. Damping ratios, influenced by material properties and connection types, provide additional information; an increase can indicate internal friction due to cracking, while a decrease may suggest weakened connections.
The methodology combines computer vision algorithms for object tracking, displacement analysis, and magnification of subtle motions to determine modal parameters accurately. Signal processing techniques applied to vibration and motion data extracted from video footage allow the calculation of natural frequencies and damping ratios of field structures. Enhanced motion analysis through video magnification highlights subtle movements that traditional monitoring methods might miss, revealing indicators of abnormal behavior and advancing our ability to predict structural issues.
This non-invasive approach offers significant benefits by enabling near real-time, continuous assessment of infrastructure health. The use of videos, coupled with AI's analytical power, expands data collection to areas beyond the reach of physical sensors, enriching datasets and improving the accuracy of structural health monitoring.