On the Combination of Dynamic Features into 3D City Models

Abstract

The rapidly developing field of digital twins for different domains demands not only the expert work of people but also data and their incorporation into existing models of reality. Another issue is adding changing objects in time and space – dynamic features, into the models of interest. This paper presents a combination of dynamic features related to traffic and pedestrians and a static 3D city model and visualisation of their combination in a user-friendly form. Dynamic features representing people and cars are derived by neural image recognition from a static camera's short video recording of a particular area. Not only recognition of objects of interest but also their transformation from image coordinates into coordinates in a geographic coordinate system was performed to visualise such changing objects in a 3D city model of a particular study area in the city of Pilsen, Czech Republic. Furthermore, such features were visualised in a web map application presenting dynamically changing objects in space and time representing people and cars, together with static 3D models of surrounding buildings. This work can provide valuable information for further exploitation of data about spatiotemporal aspects of mobility in cities and a simple yet effective representation of dynamic features within 3D cartographic representation accessible via a web page.