Computer Vision Approach for Investigation on Synchronization of Bouncing Crowd

The increasing number of human-induced vibration problems for long-span engineering structures indicates the importance as well as a lack of research for crowd load modelling.The synchronization between individuals is the key for crowd load modelling.To get over the problems of considerable expense and acquiring synchronized time-history data faced by traditional experimental methods, computer vision techniques were introduced by interdisciplinary research.Firstly, the optical flow method and its application theory were investigated.A reconstruction procedure for bouncing load was developed and verified through an experiment in a gait laboratory.Crowd bouncing experiments were carried out.The synchronized time history of bouncing loads for every individual in different situations were acquired through the developed procedure.The coordination factors were defined and calculated accordingly.The values of coordination factors at different frequencies of metronome beats and numbers of people were given and investigated.The analysis results show that the method and the procedure developed are capable for reconstructing crowd-bouncing loads and evaluating the synchronization of a bouncing crowd quantificationally.