Parallel Algorithms For Remote-sensing Image And Its Implementation Of Grid Services

As a synthetic subject, remote sensing has gotten rapid developement; nowadays it plays a more and more important role in civil and military applications. With the development of remote sensors, remote platforms, and data communication technologies, remote-sensing data is increasing in an incredible speed. Meanwhile, fast processing technologies for many domains become more and more urgent.Recently, fast processing for remote-sensing image is one of the focus researches at home and abroad, the key to solve the problem is using parallel processing. However, many problems to be solved include not only parallel algorithms for remote-sensing image, but also how to confront with the new puzzles of the mass remote-sensing data and how to make the applications of these parallel algorithms better.We analyse these requirements in this paper, aiming at the research and implementation of parallel algorithms for remote-sensing image processing, also at the implementation of the services for these parallel algorithms in grid enviroments. The contributions and relevant work in the paper are as follows:Firstly, a parallel geometric correction algorithm based on dynamic division-point computing is presented. We have analysed some existed parallel geometric correction algorithms for their limitations. This new algorithm has good data locality, better load-balancing, and the globe computing is less.Secondly, a new smooth method to solve the excessive segmentation with which watershed segmentation faced is presented. As we implement parallel watershed segmentation algorithm based on OQ, a new smooth operator is presented. This algorithm can solve the excessive segmentation which watershed inherents better.Thirdly, an algorithm for parallel embedded zerotree wavelet encoder is presented. We have researched embedded zerotree wavelet encoder algorithm and analyed its parallism. After that, a parallel algorithm for EZW based on image division PEZW-ID is presented.Finally, we implement PRIPP package and deploy it in grid enviroment. We introduce the architecture of PRIPP package and design framework, then we implement and deploy it as Grid Service: PRIPS-G, in a special grid enviroment. The interfaces and results have been given in this paper.The algorithms we have studied were analysed in academic modes and tested in YH MPP, the Grid Services of PRIPS-G is applied in ChinaGrid Project.