OSMsc: a framework for semantic 3D city modeling using OpenStreetMap

Abstract

Semantic 3D city models have been widely used in computer graphics, geomatics, planning, construction, and urban simulation. While traditional geometric models are used only for visualization purposes, semantic 3D city models contain abundant detailed information, such as location, classification, and functional aspects. Such semantics can facilitate a better interpretation of the built environment by computers. However, the current semantic 3D city models are mostly specific to particular city object types and features, with unclear spatial semantics, which limits their broader applications. This study, therefore, proposes a novel framework called OSMsc, where OSM refers to OpenStreetMap and sc refers to semantic city. The OSMsc framework considers OSM as the primary data source to construct city objects within the specified study area, construct semantic connectors, enrich spatial semantics, and generate the CityJSON-formatted model. The case studies demonstrate that semantic 3D city models constructed by OSMsc are free from geometric and semantic errors, applicable to any city worldwide, and have potential for urban studies, such as urban morphology and urban microclimate analysis.