Research On The Optimization Method For The Organization And Scheduling Of Massive Topographic Data

Terrain is the place where we live.The exploration of the terrain helps us to better understand the environment in which we live.In recent years,with the continuous development of the digital earth,people's simulation of the earth's surface has changed from two-dimensional to three-dimensional.3D terrain visualization technology is also the focus of research in the fields of computer,GIS,virtual reality,etc.With the expansion of research areas and the improvement of terrain resolution,the amount of terrain data that needs real-time processing is very large,bringing tremendous pressure on limited computer memory and processing power.In order to solve this problem,domestic and foreign scholars have done a lot of research and obtained many important research results,but there are still some unresolved problems,For example,multi-level terrain data storage has a large amount of data redundancy,terrain data encoding methods are not perfect,and data prefetching strategies are not efficient.Therefore,in order to realize real-time visualization of massive terrain data,it is necessary to effectively solve the problem of computer storage,organization and scheduling of large-scale terrain data.In view of the above problems,this paper starts from the aspects of multi-level terrain data non-redundant storage and coding,efficient terrain data prefetching(cache)strategy,optimizes existing related algorithms,and designs and develops corresponding experimental systems for this paper.The optimization method is verified.The main contents are as follows:(1)An index structure based on layer,order,row and column is proposed to improve the retrieval efficiency of terrain data of different levels.In view of the problem of data redundancy and coding incomplete multi-level storage of massive terrain data,the traditional quadtree data structure is used to hierarchically block the terrain data.When the database stores data,it is not in the high-level terrain block.The data stored in the low-level terrain block is stored,and only the data increased due to the increase of the resolution is stored,thereby eliminating the redundancy of the data;then,for the above-mentioned non-redundant data storage method,an efficient large-scale terrain is proposed.The data storage and encoding index method improves the retrieval efficiency of different levels of terrain data.The experimental results show that the multi-level terrain non-redundant storage method optimized in this paper can reduce data redundancy to a large extent and maintain high data retrieval efficiency.(2)Improvement of data prefetch scheduling strategy.Aiming at the problem of insufficient prefetch cache obtained by using the node bounding box,on the basis of the original trapezoidal domain clipping,a circular prefetching strategy is added,and data loading and unloading is realized in combination with multithreading.First,use the spherical(circular projection)field of view to intersect the node to surround the ball,as the prefetching area of the data.Then,using the improved fan-shaped projection field of view and the terrain block node to surround the ball to intersect,as the drawing area of the data.The experimental results show that the spherical field of view proposed in this paper can meet the needs of data prefetching to a large extent,and the scheduling efficiency is higher than the traditional method.(3)Design and develop a massive terrain visualization experiment system,which uses different scales of SRTM3 terrain data to compare and verify the optimization method proposed in this paper,and the scene roaming function is implemented in the system.The experimental results show that with the optimization method proposed in this paper,the system can run smoothly at a high frame rate and achieve efficient and real-time interaction.