Automatic Feature Extraction From Raster Nautical Charts Using A Constraint Programmed Expert System

David Pendleton, Shin-yi Hsu and Jane H. Hsu

Abstract:	Efforts to automate chart production systems have been attempted within the Federal Government over the past decade. Civilian and military organizations alike have been challenged in their efforts to use computer technology to increase productivity and reduce costs. The most important limiting factor to such efforts is the expense of creating, loading, and maintaining the large vector data bases required to store chart feature attributes and graphic information. The Office of the Coast Survey, NOAA, currently uses raster nautical chart production techniques in lieu of the vector database approach. This method will be extended to combine the advantages of both raster and vector products by implementing a new hybrid raster/vector chart production system. The hybrid will reduce the volume of vector data to be collected, loaded, and maintained to only those features critical to navigation. The raster base will add the remaining features for consistency with the current chart form. To reduce data collection and maintenance costs, new methods are needed to eliminate, where possible, tedious and expensive manual digitizing operations. This paper summarizes the results of a CSDL software prototype which uses an expert system to recognize sets of raster chart features, perform semantic interpretation, and convert the resulting information into vector form. The project is under contract to SRE, Inc. using their IMaG system to achieve automatic feature recognition and classification through an iterative feedback training process. IMaG integrates Image processing, Multisource analysis and GIS into a single environment. IMaG has successfully extracted typical feature sets from raster NOAA charts and has demonstrated its feasibility to overcome the limitations of conventional technologies. Future research will finish the task by converting the extracted information into error-free attributed vector representations.
Published in:	Proceedings of the 53rd Annual Meeting of The Institute of Navigation (1997) June 30 - 2, 1997 Albuquerque, NM
Pages:	585 - 593
Cite this article:	Pendleton, David, Hsu, Shin-yi, Hsu, Jane H., "Automatic Feature Extraction From Raster Nautical Charts Using A Constraint Programmed Expert System," Proceedings of the 53rd Annual Meeting of The Institute of Navigation (1997), Albuquerque, NM, June 1997, pp. 585-593.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In