A Novel Approach for Generation and Visualization of Virtual 3D City Model Using Open Source Libraries

Abstract

Virtual 3D city models are digital representation of terrain surfaces, sites, buildings, roads, waterways, etc. In general, these models are used for presenting, exploring and analyzing the urban areas. These models are also used in domains of smart city, virtual reality, infrastructure planning, telecommunication, disaster management, real estate services, education, tourism and change detection. There are various approaches and commercial packages to generate virtual 3D city models. In this novel approach, we have not used any commercial package or tool toward generation and visualization of city model but instead developed in-house routines and used libraries from open source domain to generate and visualize real-time virtual 3D city models. Generation of 3D city models involves the imagery from high-resolution satellite, high-resolution digital elevation model (DEM) and vector shape files from open street maps (OSM) (https://www.geofabrik.de/geofabrik/openstreetmap.html) as inputs. Over and above virtual 3D city models, OSM vector layers of road network, river network and important places have been overlaid for making the virtual city model more informative. Toward realism, virtual trees are placed in the open spaces (as per shape file markings), Sun position and day/night effects, ambient condition effects and building texture, has been incorporated. To achieve generation of city model, we mosaicked preprocessed DEM scenes and images, resampled given datasets using cubic-resampling algorithm, registered the image, DEM and vector layers using phase correlation algorithm, and extracted the height information by processing of DEM and shape files. Vector layers are registered using latitude and longitude and overlaid over the 3D city model for information about the places, roads and water bodies. For efficient visualization and rendering of the model, open scene graph library and tiling-based approach are used in which tiles are replaced in the main memory by its sub-tiles based on viewer’s position. Mentioned approach is applicable to generate and visualize city model of any given area provided the DEM, imagery and OSM data. This paper mainly discusses the approach and advantages of this new cost-effective approach to develop virtual 3D city models using Indian Remote Sensing datasets.