期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2010
卷号：XXXVIII Part 4
出版社：Copernicus Publications
摘要：With the fast development of information technologies, nowadays the collection of people flow data becomes much easier andwe can have different kinds of measurement data, such as train use data gotten by IC card, high way use data gotten byElectronic Toll Collection System, and so on. However, most of them have been used separately. In this research, we are tryingto combine these different kinds of observation data together to make a more accurate estimation about people, based on dataassimilation techniques. We propose an algorithm using Particle Filters for data assimilation of people flow data and estimatetrajectories with it, assuming that we can get the number of people who pass each detecting line as observations. In thisalgorithm, we make particles when getting an observation that a person enters a detecting line and evaluate when getting anobservation that a person goes out a detecting line. Particles are made by a probabilistic model built by trajectory data gotten bythen. We first apply this algorithm to a simple example to consider the effectiveness. Based on this result, we then apply it tothe actual trajectory data, pedestrians’ flow data gotten in an area of about 60 × 20 m2 at Osaki station in Japan. This algorithmis justified by the combinations of orientation and destination, walking times of each person, and trajectories. In order to validatethis algorithm, we use trajectory data that is measured by laser sensor at Osaki station.
