Research And Design Of Distributed File System Based On Multi-resolution Spatial Image

Efficient storage and quick query of multi-resolution spatial image are very important to the large-scale terrain visualization application.In this thesis,a multiresolution spatial image distributed file system MR-Ceph(multi-resolution Ceph)is designed to store the multi-resolution spatial image according to the geospatial index RPyTree and Ceph distributed file system.The RPyTree index which consists of R-tree and image pyramid index can speed up the query of multi-resolution spatial image.In order to ensure the consistency of the spatial index on multiple service nodes,method of version control and timed update is established.When MR-Ceph does not contain the queried image with a specific resolution,it is necessary to obtain the image of the specified resolution by resampling.To accelerate the resampling speed on a single node of this distributed file system,a RECG model is proposed to choose a CPU or GPU resampling method.Experimental results show that MR-Ceph can effectively manage multi-resolution spatial images.