fferent land covers can present very similar spectral features (de Jong, S. M., 1993). This interclass confusion introduces errors into the resulting spectral classification.
Geographic Information Systems (GIS), have developed from three largely separate origins. The Canada Geographic Information System (CGIS) in 1963, vector based digital mapping in the 1960s, and the Computer Aided Design (CAD) area. GIS systems have grown from digital mapping systems by the addition of a relational database for the storage of attribute data and analysis functionality, and it is this vector data model that has become dominant in GIS (Jones, Settle, and Wyatt, 1988).
GIS systems from each of these three origins have evolved towards a common basic range of capabilities input, display and manipulation of both vector and raster data, raster DTM handling, topological structuring of vector data, vector and/or raster polygon modeling and analysis, and storage of attribute data in a relational database management system. Most systems today have the ability to perform basic image handling tasks, such as rectification and display as an image backdrop. Some systems allow an image processing package to share the same screen as the GIS, giving access to both image processing and GIS functionality. In addition, modern GIS system architectures can handle large raster and vector datasets in a seamless manner, so image processing can escape the constraints of image scenes and map data can escape the fetters of sheet boundaries (Faust, Anderson, and Star, 1991).
In visual classification, an interpreter evaluates several image characteristics (e.g., tone or color, texture, size, shadow, pattern, location and associations) in order to identify and deduce the significance of the components of the image. The majority of those characteristics are not used in conventional digital classifications.
Attempts using different approaches have been mad...