结合MapReduce和HBase的遥感图像并行分布式查询

Parallel and Distributed Retrieval of Remote Sensing Image Using HBase and MapReduce

提出了一种可扩展的遥感图像多维度并行查询模式,即利用MapReduce实现海量图像数据金字塔的并行构建,利用HBase实现图像的分布式检索,设计和实现了单张遥感图像金字塔的并行构建方法和图像索引系统。实验结果表明,随着Hadoop和HBase集群的增长,图像数据的导入和检索速度得到明显提升。

With the development of satellite remote sensing technology, the volume of remote sensing image data grows exponen- tially, while the processing capability of common computer system is hard to satisfy the requirements of remote sensing image data accessing and retrieval. In this paper, we propose a scalable and parallel processing model based on MapReduce and HBase mechanism. It is a distributed and parallel storage method which combined with Pyramid Model and MapReduce Thinking. It recodes the tiles of each remote sensing image and defines the storage rule to ensure the tiles can be stored and searched in parallel. Experiments show that the speeds of data importing and data retrieval increase obviously as the cluster of Hadoop and HBase grows.