A High-Performance Spatial Range Query-Based Data Discovery Method on Massive Remote Sensing Data via Adaptive Geographic Meshing and Coding

Abstract

Many countries and organizations have been sharing a vast amount of remote sensing (RS) images across the Internet via their portals to promote the flourishing development of RS applications. These portals provide the data discovery ability to the public with necessary ways. Spatial range query is one of them to enable data discovery on RS images by their spatial extents. This article first analyzes the technical challenges in building a spatial index based on the key-value model. Then, this article presents how our proposed adaptive geographic meshing and coding method solves these challenges. The proposed method transforms the spatial extent of RS images to key-value-based indexing records to ensure spatial proximity as well as a low ratio of data duplication in indexing records. This article also shows how the proposed query selectivity strategy improves the performance of spatial range queries on indexing records. The strategy determines the order of two essential operations in a spatial range query, namely, the data deduplication and the spatial relationship computation. Finally, we implement a prototype system based on HBase and develop two MapReduce-based algorithms to speedup the process of the index construction and the spatial range query. Extensive experiments on a real dataset demonstrate that our proposed method owns a better query performance as well as a lower space overhead than competitors to handle spatial range queries on RS images.