The Discovery Of The Basic Terrain Features Based On Integration Technology Between GIS And Spatial Data Mining

The rapid development of data collection technology brings up "The sea of data". Increasingly abundant spatial data and attribute data is collected and stored in the spatial databases,but it is already far beyond the human's comprehensive faculty.The current GIS spatial databases system can efficiently complete such functions as the read-write, modification, statistics,query and so on, but the implicit and obscure knowledge in spatial databases can not be discovered,such as spatial relationships,rules and develop trends,which results in the present condition of "data drowning but starving for knowledge". It is an urgent problem to how to extract useful information quickly from spatial databases.By means of taking data mining technology into GIS spatial databases,the implicit,latent and useful spatial features and patterns inside GIS spatial databases can be discovered.So the integration of GIS with spatial data mining technology is very necessary.Firstly,the integration technology can advance GIS query and analysis function to the new stage of discovering knowledge.Secondly, the information discovered from GIS spatial databases may constitute the knowledge bases,which can be used to establish the intelligent GIS system.The computer discovery technology of terrain features in topographic map is a new research project at home and abroad .We analysed the terrain features with the DEM before. With regard to the DEM disadvantages of bulk storage capability, the low accuracy and the great error, in this paper the author uses CD-TIN which is constructed on the basis of contour lines to discover and extract terrain features.After synthetically analysing three classical algorithms of constructing D-TIN,The author uses improved convex algorithm to construct non-constrained D-TIN,and then puts forward impacted-region-diagonals-change local adjusting algorithm.and the new algorithm is used to embed contour lines into D-TIN as constrained segments and ultimately construct CD-TIN.Delaunay triangles which take contour lines as constrained segments can't span contour lines. We can discover flats and mountanious regions inside topographic map by means of computed normal vector and gradient.In the discovered mountanious regions,terrain features such as the neks,mountain peaks and valleys can be also discovered through the established CD-TIN's data structure and topology relationships.The paper firstly introduces the origin of data mining knowledge,the development of spatial data mining and the content of the paper.Thetheoretical knowledge of data mining is narrated at Chapter 2 in detai.At Chapter 3,the author describes the spatial data mining technology and analyses the necessity of the integration technology of GIS and spatial data mining technology.The author uses the constructed CD-TIN to discover and extract the terrain features.So after three classical algorithms of constructing D-TIN are reviewed,the author uses improved convex algorithm to construct D-TIN and puts forward impacted-region-diagonals-change local adjusting algorithm to construct CD-TIN.At Chapter 5 the terrain features is discovered by means of the constructed CD-TIN.At the end of the paper, the author summarizes the work and proposes the future of spatial data mining of GIS.