面向Spark的遥感影像金字塔模型的并行构建方法

摘要

随着遥感技术和摄影测绘的发展,遥感影像的分辨率不断提高,数据量日益增长,这对快速、高效地处理海量遥感影像数据提出了更高的要求,如何有效、智能地存储和处理海量遥感数据成为研究的热点.在分析现有金字塔模型的并行构建的基础上,设计一种面向Spark计算框架的影像金字塔模型.模型给出了影像金字塔构建算法及影像数据的分布式存储组织结构,实现了海量遥感影像数据在Spark中的并行处理,为Spark增加了计算处理空间数据格式的能力.实验结果表明,利用该方法能够在Spark云平台上实现快速、高效的解决海量遥感影像金字塔的并行构建,特别是在面对海量遥感影像数据时,无论从金字塔构建性能上还是遥感影像的计算效率上,Spark都更具优势.%With the development of remote sensing technology and photogrammetry, the resolution of remote sensing image is increasing and the data is growing day by day.This has put forward higher requirements for fast and efficient processing of massive remote sensing image data, and how to store and process effectively and intelligently remote sensing data become the research hotspot.On the basis of analyzing the parallel construction of existing pyramid model, a Spark image pyramid model is designed.The model provides the image pyramid construction algorithm and the distributed storage structure of image data.The parallel processing of massive remote sensing image data in Spark is realized, and added with the ability of calculating and processing the spatial data format for Spark.The experimental results show that this method can achieve fast and efficient solution to the parallel construction of massive remote sensing image pyramid on Spark cloud platform, especially in the face of massive remote sensing image data, both from the pyramid to build performance or remote sensing images on the calculation efficiency, Spark is more advantages.