Study And Implementation Of Remote-sensing Image Parallel Processing System Based On CGSP Under IPv6

Remote sensing information is an important foundational resource, and various trades are widespread to apply remote sensing images. But, along with the remote sensing technology's development and mature, the remote sensing image's scale increases, the imagery processing's real-time request grows, the remote sensing resource's sharing demand is more and more urgent. However, traditional processing methods are unable to satisfy the applications. It is an effective way solved the questions to combine the parallel processing technology with the grid technology.Therefore, in this thesis we researched on the parallel processing technology and the grid service implementation of remote sensing images faced grid situation, and designed a remote-sensing image parallel processing system based on CGSP, which intergraded with IPv6 and grid technology. The main contributions of this thesis are as follows:1. We optimized the parallel PCA fusion algorithm for grid. The PCA algorithm is a classical fusion method, which application is extremely widespread. But, the direct parallel method's communication time is too many, which has restricted the operating speed's enhancement seriously. We designed and implemented a kind of parallel PCA algorithm for grid called PPCA-G. It is based on polynomial transform aiming to reduce communication cost and improve the efficiency compared with the direct parallel method. The experimental results show that this algorithm achieves a good speedup ratio, extensibility and good degree of parallelism.2. We designed and implemented a remote-sensing image parallel processing system. We fully used the technical superiority which IPv6 and CGSP have, to design a remote-sensing image parallel processing system under IPv6 based on CGSP. We used the hierarchical modular design method during the implement, which is extremely advantageous to the software's maintenance and expansion.3. We researched the implement method about to seal a remote sensing image parallel processing algorithm to be a service. We expatiated the system configure and the implement step, and made the improvement to the original service executive in the thesis. Based on them, we integrated 41 kinds of remote sensing image parallel processing algorithm to RIPPS and gave the experimental result to evaluate it.