基于不同插值算法生成DEM及对地形因子影响的研究

发布时间：2018-10-22 19:59

【摘要】：目前,DEM广泛应用于众多领域,较为常用的使用地形图数据进行构建。不同的使用目的、数据来源、地貌类型,其所构建的DEM呈现出一定的差异性,然而并没有一种通用的插值算法满足实际需求,如何选择合适的插值算法以及相应的参数就显得比较重要,不同算法所构建的DEM派生地形因子对质量评定也起着重要作用。随着数据量的增大,DEM构建速度就显得相对较慢,人们常借助于购买高性能机器,来提高数据处理的速度,但这在一定程度上无法满足对实际的需求。随着GPU技术发展,CUDA拥有其独特的加速效果,这样就可以利用CUDA技术进行加速构建DEM。本文选用福建省三种典型地貌类型样区的数据,从DEM构建精度、速度、派生地形因子质量角度进行研究,对常用内插算法的构建参数以及误差进行了对比,然后基于CUDA技术进行并行设计构建DEM,结果表明所设计的算法得到了一定程度的加速。本实验的主要内容如下：(1)不同插值算法参数对构建DEM的影响。利用质量评定指标,探讨不同插值算法参数的误差变化特点,找出不同插值算法构建高精度DEM的合理参数。在此基础上,对不同插值构建DEM的误差特征进行分析,得出ANUDEM、NN、TIN算法构建的精度较高,而IDW、Kriging算法误差相对较大。(2)基于CUDA进行并行插值算法设计以及性能评价。通过CUDA多线程技术,利用分块和动态点搜索方法,设计和实现IDW、Spline、Kriging并行算法,并用于快速构建DEM。通过对比分析三种算法的性能,三种插值算法在不同数据量情况下,呈现不等的加速。(3)探讨不同插值算法对DEM地形因子的影响。利用数理统计、信息论知识,分析不同插值算法所派生地形因子的差异性,了解不同算法提取地形因子的适用范围。结果表明AUNDEM算法提取坡向地形因子不具有平坦区域,IDW、kriging算法提取的地形因子质量较差。本实验为可为普通用户在利用插值算法对数据进行处理以及进行数字分析提供一些参考性依据,同时也可利用COM组件技术对并插值进行封装,以方便其他人员使用,具有一定的理论和应用价值。
[Abstract]:At present, DEM is widely used in many fields, more commonly used topographic map data to build. Different use purposes, data sources, geomorphological types, the DEM constructed by it presents some differences, but there is no universal interpolation algorithm to meet the actual needs, How to select the appropriate interpolation algorithm and the corresponding parameters is very important, and the DEM derived terrain factors constructed by different algorithms also play an important role in quality evaluation. With the increase of data volume, the speed of DEM construction is relatively slow. People often use high performance machines to improve the speed of data processing, but to a certain extent, it can not meet the actual needs. With the development of GPU technology, CUDA has its unique acceleration effect, so that CUDA technology can be used to accelerate the construction of DEM.. In this paper, the data of three typical geomorphological types in Fujian Province are selected, and the construction parameters and errors of common interpolation algorithms are compared from the angles of DEM construction accuracy, speed and derived terrain factor quality. Then the parallel design based on CUDA technology is used to construct DEM,. The results show that the proposed algorithm is accelerated to some extent. The main contents of this experiment are as follows: (1) the effect of different interpolation parameters on the construction of DEM. Using the quality evaluation index, the error characteristics of different interpolation algorithm parameters are discussed, and the reasonable parameters for constructing high precision DEM by different interpolation algorithms are found out. On this basis, the error characteristics of different interpolation to construct DEM are analyzed, it is concluded that the accuracy of ANUDEM,NN,TIN algorithm is higher, while the error of IDW,Kriging algorithm is relatively large. (2) parallel interpolation algorithm design and performance evaluation based on CUDA. The parallel algorithm of IDW,Spline,Kriging is designed and implemented by using CUDA multithreading technology, using the method of block division and dynamic point search, and it is used to build DEM. quickly. By comparing and analyzing the performance of the three algorithms, the three interpolation algorithms show unequal acceleration under different data volume. (3) the influence of different interpolation algorithms on the terrain factors of DEM is discussed. By means of mathematical statistics and information theory, the differences of terrain factors derived from different interpolation algorithms are analyzed, and the applicable range of different algorithms to extract terrain factors is understood. The results show that the AUNDEM algorithm does not have a flat region for extracting slope terrain factors, and the quality of terrain factors extracted by IDW,kriging algorithm is poor. This experiment can provide some reference basis for ordinary users to process and analyze data by interpolation algorithm, and can also encapsulate and interpolate the data by using COM component technology, so as to facilitate other people to use it. It has certain theory and application value.
【学位授予单位】：福州大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P208

【相似文献】