Research On Parallization Of Projection Transformation Algorithm Based On PRPL For Raster Map

Map projection transformation is a basic research topic in the field of geography and geographic information science research,and it is also an important link that affects the quality of map production.The map projection transformation includes vector map projection transformation and raster map projection transformation,and the vector map projection transformation is mature enough at present.With the advantage of simple raster data structure,convenient stack analysis,surface simulation,and strong data acquisition,the grid map has become one of the standard map production.However,the grid data includes the mathematical basis and the geospatial position,are characterized by large amount of data,so any transformation of the grip data will affect the grip map quality after projection.Therefore,the research on the accuracy and efficiency of grid map projection transformation is still the basic research problems in the field of geography and geography.On the issue of huge data and low efficiency of projection,based on pRPL(Parallel Raster Processing Programming Library),this paper focuses on the parallelization of the algorithm for raster map projection transformation.The main contents of the thesis include:(1)Based on the process of the raster map projection transformation,this paper summarize the implementation of the raster map projection transformation algorithm and analysis the mathematical principles of the relationship between the coordinates of the transformation between the projection,resampling methods,etc.Finally,serial program that implements the raster map projection transformation algorithm is completed.(2)Based on the serial raster map projection transformation algorithm,this paper used performance analysis tool to locatethe time-consuming functions in the algorithm,and the parallelism of thosefunctions with the mathematical principle of the algorithm is analyzed.Based on the different load balancing and I/O mode provided by pRPL,raster map projection transformation algorithm is designed and implemented separately.(3)The parallel implementation and test of the parallel algorithm on the multi-core platform,which is used to find the optimal load balancing and I/O mode for raster map projection transformation algorithm in pRPL.In the optimal mode,the time consumed by various parts of the program is analyzed,and the reasons for the efficiency of the processing are found.The simple parallel tiff writer(SPTW)is used to write the output file to optimize the parallel algorithm.The optimized algorithm is compared with a tool,Librasterblaster,for converting raster map projection on the same platform.The experimental results show that the parallel algorithm based on pRPL can improve the efficiency of the algorithm.The optimization algorithm based on the initial parallelism can further improve the efficiency of the algorithm.Compared with Librasterblaster,the acceleration ratio of this algorithm is obviously promoted.In this paper,this paper proposed a parallel algorithm based on pRPL for raster map projection transformation,which improves the efficiency of raster map projection transformation and provides a reference for the parallel implementation of the future raster processing algorithm on pRPL.It has a certain technical reference value.