Research On Remote Sensing High-performance Computing Strategy And Preliminary Implementation Of Platform Based On Cluster

Ecological crisis，such as over-exploitation and destruction of natural resources and theenvironment, has been the biggest threat to human beings. As one of the most important meansfor environmental monitoring, remote sensing technology plays an irreplaceable role inresource evaluation and testing, environmental pre-estimates and pre-warning for ecologicaldisasters. But in recent years, with the rapid development of remote sensing technology, theneed of effectiveness of remote sensing processing has deeply increased, massive multi-sourceheterogeneous remote sensing data fusion processing and the remote sensing services forsociety has become the serious problem in the area of remote sensing.Based on the problems faced by remote sensing processing, and the analysis of remotesensing data processing flow as well as remote sensing algorithm, this paper proposed a set ofremote sensing high-performance processing strategy, and designed and initial realized theremote sensing high-performance computing processing platform by the service principle,achieved three different parallel processing algorithms models and wetland vegetation typesinformation extraction automated processing based on platform.The main contributions of this paper are as following.(1)Based on the characteristics of remote sensing algorithms, this paper classified theremote sensing algorithm and summarized the general process for remote sensing dataprocessing and analysed of remote sensing process parallelism. At the same time, this paperproposed the strategy for parallel processing of remote sensing high-performance based oncluster and the corresponding parallel model for remote sensing processing such as pixelprocessing algorithm, the local processing algorithms and global processing algorithms,and atlast achieved each kind algorithm.(2) On the basis of remote sensing high-performance parallel processing strategy, wedesigned a high-performance processing of remote sensing platform architecture, as well asrelated functions by the core principles of service, and achieved the interface between the platform functional layer implementation. Initially realized of the basic functions of theplatform based on the cluster of Chinese Academy of Forestry. The remote sensing servicesplatform.can support parallel processing through servicesation of data and algorithms andcluster job scheduling and flowing of processing.(3) In the environment of cluster, we realized the basic functions of remote sensinghigh-performance processing platform and show the core business function by the example ofwetland vegetation type extraction processing. Through the implementation of the platform, itis possible to solve the efficient problem of remote sensing, and improve the speed ofprocessing the massive data based on the cluster, and provide the data and products services forusers.