基于网格环境的遥感图像并行分类  

PARALLEL CLASSIFICATION OF REMOTE SENSING IMAGES BASED ON GRID ENVIRONMENT

作　　者：张雁[1,2] 吴保国[1] 王晓辉[2] 林英[3] 

机构地区：[1]北京林业大学信息学院,北京100083 [2]西南林业大学计算机与信息学院,云南昆明650224 [3]云南大学软件学院,云南昆明650091

出　　处：《计算机应用与软件》2015年第2期194-197,219,共5页Computer Applications and Software

基　　金：国家高技术研究发展计划项目(2012AA102003)

摘　　要：随着遥感技术的发展,高分辨率大容量遥感数据的应用,对图像处理效率提出了更高的要求。网格计算因具有分布式、高性能和充分的资源共享性,为海量遥感图像的处理提供了有效的解决途径。针对遥感图像分类,提出基于网格环境的遥感影像并行模型,分析构建此模型的网格服务机制,设计网格服务及任务调度的算法流程。搭建网格实验测试平台,采用封装的SVM分类服务,实现了遥感图像并行分类处理。实验结果及分析表明,测试平台实现了网格环境下的遥感图像并行分类的架构,有效提高大容量遥感数据的分类效率,为分布式并行处理遥感图像提供了有效的途径。With the development in remote sensing technology and the application of large volumes remote sensing data with high- resolution, higher requirements have been put forward on image processing efficiency. Since grid computing has the distributed and sufficient resources sharing property with high performance, it provides effective solution approach for massive remote sensing images. In view of remote sensing classification, we present the grid environment-based parallel remote sensing image model, analyse the grid service mechanism of building this model, and design the algorithm flow of the related gird services and task scheduling. We set up test and experiment platform in grid environment, and implement the parallel classification processing on remote sensing images by employing encapsulated SVM classification service. Experimental results and analysis show that the test platform realises the architecture of parallel classification for remote sensing imagesin grid environment, which effectively improves the classification efficiency of large volume remote sensing data, and provides an effectiveway for solving the distributed parallel processing of remote sensing images.

关 键 词：网格环境 遥感图像分类 并行处理架构 网格服务 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]