In recent years, with the rapid development of information science and technolog y, space information technology has been widely applied in various industries and sect ors of the community, such as transportation, geological disaster monitoring, aerospac e, natural and cultural heritage. Spatial data is one of the main research contents of spa tial information technology, has massive, multi-source, multi-resolution and other cha racteristics. with the respecting and application of digital earth, digital city, smart plan et and smart city, put forward higher requirements for spatial data storage, transmittssi on, service and other operations. A spatial index is one of the key technologies of imp roving the performance for spatial database, which directly affect the performancee of storage efficiency and spatial retrieve for spatial data. Research on spatial indexing te chnology and better indexing of spatial data, become a hot spot in the area of spatial i nformation.This article started with the main indexing techniques in spatial database. First of all, set out fundamental principles of spatial index and the main method, and then pars ed spatial index technical features, advantages and disadvantages. Secondly, put forwa rd the direction of improvement and optimization, as well as content-based image retri eval technology for the insufficient of r-tree index. Subsequently, presented Hilbert r-t rees of clustering index algorithm based on k-means after improved, and it was impro ved on the basis of the original Hilbert r-tree. Before building the tree, extracted the fe ature using Wavelet transform on the spatial object, then use the spatial data after proc essing to build tree. At the time of building, use improved clustering index algorithm based on k-means. Finally, achieved to the preliminary application for Hilbert r-trees o f clustering index algorithm based on k-means after improved in the world heritage sy stem. Hilbert r-trees of clustering index algorithm based on k-means after improved ca n be used for massive data or a large proportion of spatial data environment, and had t he obvious query efficiency. (1) Based on spatial database and spatial indexing technology, and compared the R tree family, including R tree, R+tree, R*tree and Hilbert R tree, studied their advantage and shortage.(2) Put forward the improvement direction of spatial indexing technology, improve the spatial use efficiency of node, put forward image index based on content. It uses low frequent values of image substitute the original image, abandoned the invisible part, and reduced the size of the image.(3) Introduced wavelet transformation theory, and uses wavelet transformation theory extracted the image feature, extract the low frequent value of the image; introduced the cluster analysis increase the establish efficiency of spatial index, according to the limit of traditional K means cluster analysis in some field, based on the traditional method put forward a improved K means cluster analysis method. This method has the characteristics of self-adapted cluster number and cluster center determination, and uses the maximum distance to choose cluster center. According to evaluate the cluster number, makes the number of cluster K more reasonable and stable.(4) Due to the asymmetry of spatial objects, directly establish Hilbert R tree will create big leaf node, and produce more overlapping, and result in low efficiency of searching. In order to enhance the effects of image processing, this article attempts extract the low frequent values, combining the improved K means cluster algorithm and Hilbert R tree algorithm, add improved K means cluster algorithm in the process of Hilbert R tree index establishment, put forward Hilbert R tree based on feature extraction and improved K means cluster. The algorithm first extract the feature of asymmetrically distributed spatial object before tree establishment, according to cluster result, then manage data in a reasonable way, according to the rules of intermediate and leaf node creation, established the high efficient Hilbert R tree. This algorithm realized the divide process of dense and sparse distributed spatial objects, decreased the area of leaf node, distribution more reasonable, solved the cluster storage, reduced the overlapping of intermediate node, and increased the indexing efficiency.(5) Based on G/S model, combining feature extraction and improved K means cluster Hilbert R tree, designed world heritage display and search system. This system has powerful function, and can demonstrate world heritage effectively, and through digital technology to make people learn world heritage and facilitate the management procedure of government division, and finally enhanced the protection and use of world heritage. |