|
With the rapid development of the information society, the geographic information system (GIS) and spatial database is widely used in water conservancy, environmental, surveying, transportation, marine, land, regional planning and other fields. Spatial indexing technology as a key technology of GIS and spatial database, has now become a very important research topic in the current field. Reasonable data for the characteristics of spatial data growing sea quantify its own complexity, organization, and adapt to the efficient spatial index has become important ways and means to meet people and continuously improve the data retrieval and query requirements. First of all, on the basis of the underlying theory of the study of spatial database and spatial indexing techniques, spatial indexing techniques are widely used classical algorithm analysis and comparative study its strengths and weaknesses, and to explore the spatial index optimization and direction of improvement. Secondly, for the establishment of efficient spatial index based on cluster analysis and traditional algorithm based on an improved view of the limitations of the traditional k-means clustering algorithm emerged in some areas of application and performance deficiencies, k-means clustering algorithm. The algorithm has adaptive characteristics, to determine the number of clusters and cluster centers reasonable maximum distance method selected cluster center, based on effective evaluation criteria to determine the ideal number of clusters, so that the number of clusters k value selected more reasonable and more stable clustering results, especially for spatial data clustering. Finally, given that most of the practical problems of spatial objects usually uneven distribution of objective reality directly establish Hilbert R-tree index, the larger part of the leaf nodes, easy to produce a lot of overlap, resulting in multi-channel query affect the retrieval efficiency. K-means clustering algorithm, this paper attempts to better effect in the treatment of this class of objects the aforementioned improved integration Hilbert R tree algorithm, introduced in the process of the establishment of the Hilbert R-tree index improved k-means algorithm philosophy, Hilbert R-tree index based on improved k-means clustering algorithm. The algorithm improvement ideas achievements prior to the uneven distribution of spatial objects for effective clustering and data clustering based on reasonable organization, in accordance with the relevant rules to generate the leaf nodes and intermediate nodes, followed by the establishment of efficient Hilbert R tree. The algorithm processing sub-distribution of dense and sparse space object, so that the leaf nodes smaller area, a more reasonable distribution effectively solve clustering adjacent data storage, to a large extent reduce the intermediate nodes between overlap, and ultimately improve the performance of the index.
|