A Parallelization Software Framework For Geo-data Conversion Algorithm

With rapid development of technologies of aerospace, aviation and sense, it is in keep improving that fetching geo-data from earth, GIS face new challenge in geo-data manage, analysis, volume and real-time processing, especially for the military, the environment and disaster monitoring. It is so complexity and big for geo-data that need for geo-data processing doesn’t meet by optimization efficiency in serial algorithm and CPU. The opportunity solving above questions is the high performance parallel computing technology and many researchers in GIS has developed geo-data processing technology based grid system or high performance cluster system. It is too hard for GIS researchers to master high performance parallel computing technologies. This obstacle adds to difficulties in geo-data high performance parallel processing application. So, it is need to create a technology for parallelizing serial geo-data conversion algorithm to help GIS researchers to easily use high performance parallel computing technology.The dissertation mainly discuss technical questions that geo-data conversion parallel framework parallelization. The key algorithm and technology in the dissertation(1) Studying geo-data partition strategy. This paper research geo-data partition strategy deeply and summarize geo-data partition based on application and system. The geo-data partition based on vector data and raster data is important to design parallel algorithm.(2) Studying geo-data conversion partition strategy. Three geo-data parallel algorithm mode are concluded through the systematic analytical methods of geo-data conversion and parallel mode. And based on geo-data conversion parallel mode, the task scheduling strategy for geo-data conversion application is proposed. This kind of strategy has two basic modes:master-and-slave mode and peer-to-peer mode. Aim at different parallel mode, a variety of scheduling modes is used and combined to establish the parallel scheduling strategy of geo-data conversion algorithm.(3) Designing geo-data conversion parallel framework. After research of the geo-data partition and task scheduling strategy, the whole parallel framework of the geo-data conversion is elaborately designed and constructive diagram of all functions are drawn. By introducing the object-oriented technology, the parallel framework of geo-data conversion is implemented.(4) Application based on geo-data conversion parallel framework. Based on geo-data conversion parallel framework, this research achieves the design of VCT format conversion parallelization. The result and efficiency of the parallel algorithm is proved.Above all, the dissertation structures geo-data conversion parallel framework based on geo-data partition and task scheduling. The paper illustrates basic function of this framework, design function structure of algorithms in detail and implement this parallel framework by C++. In practice of parallelizing process of VCT conversion algorithm, the result show feasibility of the framework in running time and speed-up test.