Constructing a GIS-based 3D urban model using LiDAR and aerial photographs

Due to the increasing availability of high-resolution remotely sensed imagery and detailed terrain surface elevation models, urban planners and municipal managers can now model and visualize the urban space in three dimensions. The traditional approach to the representation of urban space is 2D planimetric maps with building footprints, facilities and road networks. Recently, a number of methods have been developed to represent true 3D urban models. Those include panoramic imaging, Virtual Reality Modeling Language (VRML), and Computer-aided Design (CAD). These methods focus on aesthetic representation, but they do not have sufficient spatial query and analytical capabilities. This research evaluates the conventional approaches to 3D urban models, and identifies their advantages and limitations; GIS functionalities have been combined with 3D urban visualization techniques to develop a GIS-based urban modeling method; The algorithms and techniques have been explored to derive urban objects and their attributes from airborne LiDAR and high-resolution imagery for constructing and visualizing 3D urban models; and 3D urban models for the Texas A&M University (TAMU) campus and downtown Houston have been implemented using the algorithms and techniques developed in this research. By adding close-range camera images and highresolution aerial photographs as the texture of urban objects, effect of photorealism visualization has been achieved for walk-through and fly-through animations. The Texas A&M University campus model and the downtown Houston model have been implemented to offer proof-of-concept, namely, to demonstrate the advantages of the GIS-based approach. These two prototype applications show that the GIS-based 3D urban modeling method, by coupling ArcGIS and MultiGen-Paradigm Site Builder 3D software, can realize the desired functionalities in georeferencing, geographical measurements, spatial query, spatial analysis, and numerical modeling in 3D visual environment.