Reconstruction of People Flow in Areas of Incomplete Data Availability

摘要

Data Assimilation is a technique that synthesizes information from a dynamic (numerical) model and observation data. To reconstruct people flow in areas that are partially invisible to sensors, we assess three data assimilation methods: Kalman filter, 3DVAR, and particle filter. While most studies focus on individual-based analysis, in this study, we process the movement of people using a dynamic continuum flow theory. We derive the dynamic model of people flow and numerically solve it using the data assimilation method. Our proposed method is validated in 1D and 2D simulation experiments and on real tracking data.