A Hybrid Tree Spatial Indexing Method Based on the Clustering Algorithm

摘要

The paper first to carry on three clustering divisions to the spatial objects using the improved TGSOM algorithm, and obtains a series of similar nodes which contain the spatial objects of which have the same quantity and the same area on the whole, as well as a few noise spots which do not belong to any similar nodes, and then constitutes the dendritic structure after removing the noise spots; afterwards, optimize for the dendritic structure using the mixed structure based on the quad-division and the R~* tree, which has eliminated the overlapping and coverage factors to the nodes, and has improved the retrieval efficiency of the mass data; Finally it has proved the validity and accuracy of the method through the experiment.