Research On Remote Sensing Image Transfer In The Distributed System

The increasing demand of remote sensing image application and sharing demands more efficient data network transfer. There are following issues in the existing image data transfer. First, the large amount of image data, especially high-resolution remote sensing image, results in the transfer of large volumes of data. Second, the required data is not necessary to be the entire image file, so the transfer which uses the whole scene data as a unit is a waste of transfer capacity. Third, due to multi-department request for data in the distributed system, servers get heavier load with the increasing clients, which results in lower transfer efficiency.In order to resolve these problems, image transfer in the distributed system is studied, based on analysis of the file transfer technique and remote sensing image data characteristic, including on-demand data transfer and network transfer architecture.On-demand data transfer aims at reducing the amount of useless transfer data and improving the utilization rate of transfer capacity. The regularity of image formats makes dynamic extraction of images possible, which gets sub-blocks of data within the scope of demand. On the basis of dynamic extraction, several sub-blocks are transferred. And these sub-blocks are combined to generate the complete result. Thus, the image data in specified area is achieved, reducing the transfer data redundancy.Network transfer architecture aims at balancing network transfer load. On the basis of file sharing technology, peer-to-peer and the distributed system, the hybrid network transfer architecture is built. Then, image data transfer strategy is proposed on the basis of file block selection strategy, node selection strategy, and the characteristics of the image transfer data organization. According to this, a multi-node image data transfer method is realized, improving the transfer efficiency.Finally, the prototype system is built, which is used to test the correctness of the on-demand transfer based on the dynamic extraction and the efficiency of the hybrid transfer architecture, verifying the research results to be effective.