并行栅格数据处理网格服务节点软件的关键技术  

Key Technology of Implementation of Computing Node Software for Parallel Processing of Raster Data

作　　者：方金云[1] 何建邦[2] 

机构地区：[1]中国科学院计算技术研究所,北京100080 [2]中国科学院地理科学与资源研究所,北京100080

出　　处：《地球信息科学》2004年第1期17-21,共5页Geo-information Science

基　　金：国家863项目2002AA114020;2001AA135210;中国科学院计算技术研究所创新项目20036020.

摘　　要：提出了栅格数据处理共享环境的概念,它包括栅格数据计算节点内部集群环境中的共享内存、共享处理器、共享硬盘、共享显示等方面的技术和计算节点间的远程图像对象调用技术。前者通过在近于操作系统层为用户提供一个便于编程、效率高的图像并行处理接口(API),屏蔽用户对操作系统层的系统调用,使用户在编程时只关心图像处理算法正确性和精度。后者通过远程对象调用技术在中间件层实现对象间的分布式计算,从而对程序员屏蔽了系统的分布性和异构性。最后介绍了基于该共享环境构建的空间信息栅格服务节点软件的体系结构,即构建空间信息栅格网格服务节点的框架。Sharing image processing environment(SIPE),which includes the technologies such as sharing memory, sharing processor, sharing hard disk, sharing display in a cluster computing node, and remote image objects innovation in computing nodes,is presented as a development/deployment environment and integration framework for distributed image processing and sharing application. It incorporates common sharing cooperation function services for developers to facilitate developing new application and integrating existing ones. The architecture of Spatial Information Grid(SIG) is also presented in this paper. SIPE also provides a reference mode and architecture that defines relationships between service components within SIPE, allowing them to seamlessly work together. Specialized SIG systems can be built by integrating domain-specific tools into SIPE to shorten development cycle and reduce cost.

关 键 词：栅格数据处理共享环境 图像并行处理 分布式对象技术 多层体系结构 空间信息栅格 

分 类 号：P237[天文地球—摄影测量与遥感]