Accelerating the Generation of Three-Dimensional Urban Models from Historical Maps

About This Grant

This project develops artificial intelligence and spatial modeling methods for the extraction of architectural and geospatial data from historical records. In particular, the project automates the process of extracting data from maps of over 10,000 American municipalities that were made by The Sanborn Map Company during the nineteenth and twentieth centuries for fire insurance companies to assess potential liability. These historical maps have value for urban and regional planning, but the use of these maps has previously been hindered by the lack of tools for efficiently extracting the data. Also, the methods developed in this project facilitate the creation of longitudinal datasets that permit further scientific research on the development of cities in the United States. The project also provides training opportunities for students and researchers. Expanding on previous methods, this project develops refined techniques for extracting data from the Sanborn maps, including building footprints, construction materials, building use (e.g., residential, commercial, industrial), and the numbers of stories. A key innovation is the advancement of methods for inferring the three-dimensional architecture and structure of buildings. This approach requires extracting additional building geometry and orientation from the Sanborn maps and ancillary high resolution orthoimagery data for extant buildings, then converting this information into graph representations for analytical processing. Using this graph representation, this project clusters buildings based on morphology to identify common building templates and uses the parameters of these templates to accelerate the reconstruction of buildings with detailed geometry. This project also advances and evaluates the use of generative artificial intelligence to further automate these steps, the validity of which can be assessed through comparisons to still existing historical buildings and structured qualitative analysis by local experts. Key outputs are disseminated in accessible formats to urban and regional planners. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.