Research And Application Of Spatial Index Algorithm Based On Hybrid Cluster Analysis

With the development of Internet and GIS(Geographical Information System),WebGIS has been becoming the hot spot. Via interacting with browser and using network technologies it realizes the query and the retrieval of graphic data and attribute data among different clients based on the spatial database. Efficiency of query and retrieval for spatial data is an important performance Indicators to measure the spatial database and GIS. As the complexity and the mass of spatial data itself, it must be use spatial data indexing techniques to achieve fast query of massive spatial data. To select efficient indexing techniques is an important method of improving the performance of spatial database and GIS.The generating process of R-Tree is a typical clustering issue. Using hybrid clustering algorithm, we make the adjacent object or the closed object cluster and linear ranking with the Hilbert Fractal curve. Based on this, a Hilbert R-Tree is generated. According to this idea, a high performance spatial index algorithm based on hybrid spatial clustering algorithms is proposed. It is shown that although the algorithm increases storage spending, it reduces response time of spatial query, has the faster search speed and improves the performance of query.The main works in the paper are introduced as following:Firstly, we detailly analysis R-Tree and the struction, the construction principle and the query process of improved algorithm of R-Tree.Especially, the key study is the divided formula which is caused by adding new spatial objects and the applicable scope and the questions of this formula.Secondly, we analysis the defects of K-means clustering algorithm. After expanding and improving the traditional clustering algorithm, the hybrid clustering algorithms for the Hilbert R-Tree is proposed. Thirdly, we propose dealing with the accumulated spatial and the rare spatial separately.Fourthly, the hybrid clustering algorithm is in application to the generation process of R-Tree and the spatial index algorithm based on hybrid clustering is proposed. In the generating process of R-Tree, we use the idea of hybrid cluster which clusters the adjacent object or the closed object form R-Tree.It gains smaller coverage of nodes area than the average R-Tree algorithm, and has a higher retrieval efficiency and almost the same contribution time of the R-Tree, so it has higher storage rate. Finally, we give the flow of spatial index algorithm based on hybrid clustering.At last, it gives experimental data and analysis of the spatial index algorithm based on hybrid clustering,and applies the result to the hydrology data management system of DongGuan City.