A constraint satisfaction approach to geospatial reasoning

  • Authors:
  • Martin Michalowski;Craig A. Knoblock

  • Affiliations:
  • University of Southern California, Information Sciences Institute, Marina del Rey, CA;University of Southern California, Information Sciences Institute, Marina del Rey, CA

  • Venue:
  • AAAI'05 Proceedings of the 20th national conference on Artificial intelligence - Volume 1
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

The large number of data sources on the Internet can be used to augment and verify the accuracy of geospatial sources, such as gazetteers and annotated satellite imagery. Data sources such as satellite imagery, maps, gazetteers and vector data have been traditionally used in geographic infonnation systems (GIS), but nontraditional geospatial data, such as online phone books and property records are more difficult to relate to imagery. In this paper, we present a novel approach to combining extracted information from imagery, road vector data, and online data sources. We represent the problem of identifying buildings in satellite images as a constraint satisfing problem (CSP) and use constraint programming to solve it. We apply this technique to real-world data sources in EI Segundo, CA and our experimental evaluation shows how this approach can accurately identify buildings when provided with both traditional and nontraditional data sources.