Research On Raster Data Management Based On Ingres Relational Database

The development of computer technology, GIS technology, remote sensing technology and digital photogrammetry provides a bright future for the application of DEM (Digital Elevation Model) and remote sensing image. The vector data model has been supported by standard specifications and implemented in spatial database. However, the raster data model for spatial database storage still needs a further research and specifications. At present, most of the commercial GIS software for processing raster image files hardly meets the needs of practical application.Raster data management has three primary models: management based on the file system, spatial data engine and extended relational database. At present, Oracle, SQL Server, Informix and other large database manufacturers mainly put their research focus in the extension of relational database. One way of the extension is to add new types of data. Ingres has been developed to the version Ingres2006. It has made great progresses in functions, architecture and processing spatial data. The purpose of this thesis is to make Ingres be capable to manage raster data by extending raster data model.In response to the disfigurement of Ingres on raster data management and after deep research of raster data management mode, the thesis first designed and implemented the architecture of raster management based on Ingres RDBMS. Then, key technologies in managing raster data were comprehensively analyzed and studied. By using the Object Management Extensions (OME) provided by Ingres RDBMS, the physical storage model for raster data management, new added raster data types, together with the related raster table structure were implemented. Finally, a prototype system for raster data management based on Ingres RDBMS was built, in which storage of raster data was realized with references to the physical storage structure of raster data in Oracle GeoRaster. In order to enhance the efficiency of storage and retrieval, the prototype system stores raster data based on band and block, builds raster pyramids and establishes a spatial ordering for the raster data blocks. The performance of the prototype system was also tested and analyzed. The result showed that it was superior to Oracle GeoRaster.