Designing and Building a Vector Feature Database
Abstract
High-resolution imagery can be stored on the computer in digital form as a picture, for example, a digital raster map image file. These images then can be geo-registered by computing coefficients from points with known latitude and longitude locations. Features such as roads and airports can be extracted by applying image-processing techniques to the geo-registered raster image. Attributes describing these features and their geographical locations are stored in a "vector feature database." The vector feature database contains many feature types and is considered accurate to a given map scale. In a realtime processing system there is a need to input attributes and their locations and subsequently retrieve such feature attributes from the database with minimum processing time. The overall size of the database is also a consideration. This paper explores the design and construction of a vector feature database to 1) optimize the size of the database by reducing the number of attributes while still maintaining an adequate and unique description of the feature, and 2) enable high-speed input and retrieval of features. Several data structures that might be used to construct the database are discussed, including hash tables, binary-trees, quad-trees, and R-trees. Ultimately a quadtree structure modified to use geographic bitmaps is implemented and evaluated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 23, 2003
- Accession Number
- ADA413088
Entities
People
- Geary Layne
- Marlin Gendron
- Stephanie S. Edwards
Organizations
- United States Naval Research Laboratory