Research On Mass Remote Sensing Image Data Storage Technology

As the computer technology and sensor technology develop, the imaging resolution of remote sensing image data has been improved continuously, with increasingly enriched data types, and geometrically growing amount, which brings enormous difficulties to sharing data resources and distributing data products and new challenges to data transfer, storage and management. Based on above came this paper, aiming to resolve the key technologies in the storage, management and distribution of mass remote sensing image data.Focused on the research and discuss of mass remote sensing image data, this paper proposed a storage system named D-MaRISS (Distributed-Mass Remote-sensing Image Storage System) on the base of studying and analyzing existing storage schemes. In detail, this paper expounded the design objective and architecture of D-MaRISS, elucidated the data organization, data processing, replica strategy and fault detection & recovery of it, and analyzed several key technologies such as Cluster Management System and sensing image data organization applied in D-MaRISS.Then, this paper studied the application of distributed parallel computing in D-MaRISS, expounded the theory, programming model, process flow and fault tolerance of "MapReduce framework of parallel computing", and implemented the pretreatment of image data files and the foundation of image pyramid in D-MaRISS using MapReduce.At last, by building experimental system, this paper tested the aggregate read/write rate of D-MaRISS and the read performance of image data block. The results showed that D-MaRISS could completely satisfy the demand of this paper and efficiently sustain parallel read, which ensures its practical value and extensive applicability in managing and distributing mass remote sensing image data.