Progressive Mapping And Sampling Estimation On Distributed Vector Spatial Database Cluster

The fast development of data acquisition tools led to the explosion in the data volume,thus making scientists and engineers heavily dependent on statistical analysis and visual tools to investigate inner patterns of massive spatial information.How to handle and store the spatial data becomes an important question of geospatial data study.Meanwhile,in the practical scenario,online visualization is carried out in a progressive way,namely a sketchy view map is firstly presented,and more detailed view maps are produced gradually as the view-port scale goes deeper.One approach is to use the multi-scale spatial index technique.However,it loses the original data attribute and cannot provide spatial statistics information.In this paper,first,we study the spatial data model and investigate the data partition strategy in distributed database.The index structure of Geohash is put forward,and discusses the implementation of local index and global index in the distributed spatital database to achieve the high performance in range query.Then,based on a shared-nothing spatial database cluster system,a Geohash based data storage method is put forward.Spatial data parallel importing and distributing vector data query is realized by this data partition.Second,we provide an improved index structure,the Geo-gap tree,which aims to enhance online access to large spatial datasets,as well as enable one to compute statistical attributes like count(*)at the coarse level,and with the query continuing to execute,the query results become more and more accurate.Third,the Gap tree described suits only polygon datasets.To cope with the issues,both multi-scale spatial index and the cartographic generalization based methods was proposed as a fundamental and effective approach for dealing with real-time visualization.Its purpose is to calculate estimation value and visualization of massive vector polyline data.