Distributed Remote Sensing Image Tile Cache Method Based On Spatiotemporal And Theme Features

The remote sensing image tile service is an important port of the Web Geographic Information Service(Web GIS),which is used to implement key functions such as image retrieval and preview,image transmission and sharing,image calculation and procession,and map roaming.Improving the efficiency and real-time performance of tile service is of great significance to the construction of high-performance Web GIS.The distributed tile cache method reduces random I/O to the disk by allocating and managing hot tiles in the cluster's high-speed memory grid,which is an effective method to improve the service performance of remote sensing image tiles.However,the current distributed tile cache methods cannot combine the spatiotemporal and theme features of tiles and tile access pattern,and have problems such as pool scalability of the cache framework,unbalanced data distribution and single cache management method.In the era of big data,the current distributed cache methods cannot meet the demand of managing massive tiles and responding high concurrent requests.Therefore,this paper has studied the distributed remote sensing image tile cache method based on spatiotemporal and theme features to solve the above problems.The main contents are as follows:(1)Research on distributed remote sensing image tile elastic cache framework to improve the scalability of tile cache services.Combining the spatiotemporal features of tile pyramid model and tile service,this paper proposes a distributed remote sensing image tile elastic cache framework with high cohesion and low coupling,and a multicomponent structure and internal data streams and external data streams.The above methods reduce the degree of internal coupling of this framework.Besides,to improve the scalability of this framework,this paper proposes an elastic scaling method based on tile slot migration.(2)Research on the method of tile allocation based on spatiotemporal and theme features to improve the concurrency of the tile cache service.Combining the spatiotemporal and theme features of tile and the characteristics of tile access pattern,this paper proposes a tile partitioning method based on spatiotemporal theme coding and a tile allocation method for spatiotemporal and theme equilibrium based on the process of partitioning and then assigning.To reduce the frequency of network communication within the cluster,the former aggregates tile data with the same spatiotemporal and theme features and spatially adjacent into a single cache node.To improve the parallel response capability of the clusters,the latter separates the tile sets with the same spatiotemporal theme coding value and continuous spatial coding value between cache clusters.(3)Research on the method of tile management based on spatiotemporal and subject constraints to improve the availability of tile caching services.Combining the long-term and short-term popularity of tile access pattern,spatiotemporal locality,theme tendency,tile size difference and theme feature,this paper proposes a server-side tile cache replacement algorithm based on the spatiotemporal aging model.To improve cache hit ratio.Combining HBase(Hadoop Database),tile attributes and the spatiotemporal and theme features of tile sets,this paper proposes a method for persisting tiles based on HBase.To improve data security and tile read and write efficiency.The experimental results show that the tile partition method and allocation method proposed in this paper can reduce the average response time by about 35% and 70%,and has a high computational efficiency.The cache replacement algorithm proposed in this paper can improve the cache hit rate by about 15% while reducing the consumption of computing resources compared with the traditional cache replacement algorithms.In addition,compared with the persistence method based on relational database,the persistence method proposed in this paper can increase the read and write speed by more than 50% and reduce the disk occupancy ratio.Finally,compared with the condition of not using the cache,the cache framework proposed in this paper can reduce the average access time by more than 50%.