Vision-Based Identification of Spatio-Temporal Running Gait Characteristics
摘要
Accurate estimation of synchronization among runners in crowded environments is crucial for the vibration serviceability of slender structures, such as footbridges, which are increasingly susceptible to low-frequency loading. This contribution explores the feasibility of identifying the dominant frequency, known as the pacing rate, using overhead aerial imagery. Through color-based image segmentation and planar homography, the head trajectories of individual runners are reconstructed. These trajectories provide valuable insights into the spatial arrangement of runners and, when sufficiently precise, reveal details of the gait cycle. Analyzing these trajectories enables the extraction of spatio-temporal gait characteristics, such as running velocity and stride frequency. The findings demonstrate that the observed stride frequencies closely align with pacing rate distributions obtained from conventional methods using lower-back acceleration measurements. This non-invasive and efficient approach enables the analysis of dynamics during large-scale events, contributing to a broader understanding of social dynamics and its implications for assessing the vibration serviceability of footbridges under dynamic running actions.