Efficient R-tree index remote sensing data storage model

Abstract

The invention relates to an efficient R-tree index remote sensing data storage model. The model comprises the following specific operation steps: a file classification algorithm, establishment of a first-level index, file merging, establishment of a second-level index, and index caching. According to the R-tree index-based optimized classification tree storage model, the problems of efficient storage and rapid query of remote sensing data stored by adopting multiple distributed storage technologies are solved, the storage efficiency of the model is improved by 39% compared with that of a traditional distributed storage mode; multi-level optimized indexes are established based on an R-tree index to store index information, index caching is realized, query time is shortened, and a result shows that query efficiency is improved by 64.4% compared with a traditional distributed storage mode; the R-tree index-based optimized classification tree storage model adopts multiple distributed storage technologies to realize storage of mass remote sensing data, so that the storage problem of the remote sensing data is solved at low cost, the storage efficiency and query efficiency of the remote sensing data are improved, and the storage requirements of the remote sensing data can be met.

Description

technical field [0001] The invention relates to the technical field of remote sensing data storage, in particular to an efficient R-tree index remote sensing data storage model. Background technique [0002] With the continuous development of remote sensing technology, the storage scale of remote sensing data is also increasing. The remote sensing data of a single scene has reached the GB level and is still increasing. With the trend of hugeness, the traditional remote sensing data storage technology can no longer meet the needs of current development due to various problems such as high storage cost and poor user experience; therefore, the storage of remote sensing data is an important factor affecting the development of remote sensing technology. [0003] Using a distributed file storage system is the way to solve this problem. The distributed file system solves the storage problem of massive remote sensing data, but there is a big gap in the storage efficiency of differen...

AI Technical Summary

Problems solved by technology

[0004] The first type of traditional distributed file storage system, this type of storage system usually divides large files into small file blocks, and then stores each small file; uses Hadoop distributed file system to store spatial data, this method is based on coding technology Improve data query performance and improve data balance in HDFS; apply cloud computing technology to geographic information system, use Hadoop to realize remote sensing data storage and management system, improve remote sensing data storage management while effectively processing remote sensing data Efficiency; but HDFS is more suitable for the storage of large data files, and cannot meet the storage problem of small files in remote sensing data. When the remote sensing data contains too many small data files, it will take up too much memory and increase the master node. Burden, this type of distributed file system is suitable for the storage of a large number of large files, and the storage of small files will consume more storage space and storage access time;

[0005] The second category is a distributed storage system for massive small files. Aiming at the storage characteristics of unstructured data, a high-availability distributed storage system based on an optimized cluster MongoDB is proposed, which can not only enhance the availability of data, but also improve server performance. Scalability; etc. used MongoDB to store unstructured data for the problem of data model changes, and in order to obtain better performance, define the strategy of normalization degree and embedding degree, reduce query execution time, compared with HDFS It is said that MongoDB is more suitable for the storage of unstructured data with a small amount of data. This type of storage system is mainly designed for the storage of massive small files. It is a highly available, high-performance, and easy-to-expand distributed file system. Storage Management Solution

[0006] When processing remote sensing data, multiple distributed file storage systems are usually used. When processing remote sensing data, it is usually necessary to divide the remote sensing data of a single scene into small files containing markers, and then use the location information of the plots for these small files. Or the secondary segmentation of marker features to form smaller remote sensing data files with characteristic attributes. During the processing of remote sensing data, many small file data will be generated. Use correlation analysis to classify small files in HDFS , Merge related small files into one large file for storage, effectively reduce the load on NameNode, improve the storage and access efficiency of HDFS Shanghai’s large number of small files; people classify small files for correlation analysis, and classify the three different types of files after classification Establish different storage schemes for different types of files to achieve efficient storage and reading of small files. Establishing small file storage strategies and storing remote sensing data in different distributed storage systems can solve the problems of excessive storage space and low storage efficiency. ; Therefore, it is particularly important to develop an efficient distributed file storage model that combines large files and small files

[0007] In addition, despite the rapid development of distributed storage technology for remote sensing data, there is a lack of an appropriate method to achieve rapid storage and accurate query of remote sensing data across multiple distributed storage systems; for multiple distributed storage systems, fast and accurate retrieval Remote sensing data is also an urgent problem to be solved; the solution for storage and retrieval across multiple distributed storage systems is to build an index, retrieve the data through the index and give a response after integration, this method will generate too much index data occupation storage space, the process of integrating responses will consume too much time; et al. proposed domain indexing, this method is a new index structure category for efficiently querying data domains across multiple repositories, and domain indexing is multiple Database linkage query provides a feasible method, but for distributed storage of unstructured remote sensing data, query performance needs to be guaranteed. It clarifies how the choice of the underlying index structure affects the performance of different query operations, and shows the method used to build the index And the dynamic characteristics of the data set have a significant impact on the performance of these index structures. According to the spatial characteristics of remote sensing data and the mutuality between remote sensing data, we choose a multi-dimensional index structure: R-tree is the most popular for spatial query The index method can better reflect the advantages of multi-dimensional query in cross-distributed system storage. A new spatial clustering algorithm and r-tree insertion algorithm were proposed by et al. The comparative analysis of the spatial index performance shows that the new method greatly improves Reduced the overlap of r-tree brother nodes and kept the balance of the number of nodes

Images