Parallel Processing Technology Research And Application Based On The Cluster Of Massive Remote Sensing Data

In the past years, the technology in remote sensing field is developing rapidly. In the field of Earthobservation, with the high-resolution remote sensing satellite continually launching and more sensortechnologies such as Multispectral, hyper-spectral, microwave thermal infrared upgrading one after another,the amount of remote sensing image data obtained by remote sensing technology is increasing rapidly. Atthe same time, remote sensing applications industry has a higher demand for image data processing speedand processing accuracy."Processing difficulty" has become a growing problem. In order to improve thespeed and precision products in the remote sensing data processing stage, and solve the remote sensingdemand of environmental monitoring, disaster warning and emergency response in real time or near realtime application, in this article, Combine the characteristics in the process of remote sensing dataprocessing, with parallel computing technology, author designs and realizes a business clusteringprocessing system based on mass remote sensing data.In this paper, the main contribution and works was described as below:1. This paper researches parallel computing framework based on remote sensing data characteristics.By analyzing the characteristics of remote sensing data in the process and exploiting the concurrency in theprocess of Remote sensing data processing, also combining with computer cluster technology, this paperresearches and designs high-performance remote-sensing data cluster computing framework. In thisframework, this paper proposes Multi-layer parallel computing based on task, data and algorithm, andrealizes the design of general computing model based on remote sensing data characteristics.2. Proposed the concept of remote sensing atomic algorithm based on remote sensing algorithms, in-depth analysis of remote sensing algorithms, explore the possibility of remote sensing algorithms inparallel processing, Proposed algorithm decomposition of the original remote sensing products, Remotesensing algorithm for function-level package, defined as remote sensing atomic algorithm, develop theatomic algorithm package specification，definition of a standard atomic algorithm interface. On the basis ofthe atomic algorithm, this paper achieves processes products custom.3. This paper designs the system framework of the remote sensing data parallel processing system.After the analysis an d the research of the processes of remote sensing data processing, this paper not onlydesigns and realizes the functions of different modules, but also completes the designs of productionbusiness process. The introduction of atom algorithm improves the capability of computing greatly. Thedesign completion of this platform not only solves the problem of rapid production of remote sensingproduct, but also improves the situation of weak computing capability while computing mass remotesensing data and enhances the production efficiency greatly.