期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2000
卷号：XXXIII Part B4 (/1-3)
页码：1092-1099
出版社：Copernicus Publications
摘要：A theoretical framework for representing spatial, thematic, and temporal dimensions of geographic features has been developed. This framework relies on the unifying concept of a geographic feature as a single, unique entity in the real world with multiple object representations such as raster and vector geometries, multiple resolutions and source scales, and multiple temporal sequences. The same geographic feature can be represented in the framework at one resolution as a point and at a higher resolution as an area. Similarly, linear features such as streams can be represented as single lines at a coarse resolution and as double lines and areas at higher resolutions. Changes in spatial configuration and thematic attribution through time are also supported in the framework. Although an earlier implementation of this framework used relational database technology, implementation now focuses on object-oriented approaches. The implementation logically follows the organization of category theory, with the feature forming the basic level of categorization. An implementation for a watershed modeling application is built to test the theory. The application requires multiple feature types, such as the linear structure of the stream network embedded within the areal structure of the watershed. The multiple representations required for the watershed application include points for rain gage and stream sampling stations, which are used with digital elevation data to define polygons representing subwatersheds, vector geometry for the streams, network topology for the interstream connections and flow models, and raster geometry for the elevation surface, land cover, and soils. These requirements provide a test of the basic theoretical structure for geographic features
关键词：knowledge representation; object-oriented; multi-scale database; data models; feature extraction
