DEM分辨率确定与尺度转换方法研究

发布时间：2018-03-12 14:26

本文选题：数字高程模型　切入点：水平分辨率　出处：《南京师范大学》2014年博士论文　论文类型：学位论文

【摘要】：作为国家地理信息的基础数据,数字高程模型(Digital Elevation Model, DEM)是国家空间数据基础设施(NSDI)的框架数据,在国民经济和国防建设中具有重要的应用。在当今不同比例尺、不同分辨率、不同精度的DEM共存局面下,DEM尺度问题是急需解决的热点问题,这不仅是DEM和地学模型在集成上的根本保证,也是DEM数据推广和应用的关键所在。因此,DEM分辨率计算与转换成为DEM生产者与应用部门的重要研究命题之一。本文从原始散点数据入手,探索满足DEM地形表达要求的分辨率计算方法,当计算所得的分辨率不能与实际应用相匹配时,经常需要尺度转换。针对这一问题,本文基于分形、小波等数学理论,结合DEM的地形表达、地形分析,对DEM的分辨率计算和尺度转换进行了系统的研究,主要研究成果如下： (1)提出了基于分形理论的DEM分辨率确定方法。DEM是对地表形态的表达,应最大限度的反应地形信息量。这就需要从原始数据入手,探索DEM分辨率的确定方法。运用分形定量表达地形自相似性及复杂性的特性,建立了DEM分辨率与分形信息维数的关系,通过直线斜率差值寻求能够最大限度描述地形信息的拐点,从而确定DEM水平分辨率。 (2)研究了基于多进制小波分解的DEM尺度上推算法。从DEM的实际应用需求出发,建立了DEM尺度上推的基本原则。顾及DEM尺度上推精度因素,构建了一种基于随机数的DEM尺度上推算法。利用多进制小波的多分辨率分析、多尺度分析特性,提出了一种基于多进制小波分解的DEM尺度上推方法,通过DEM的多进制小波分解,得到的低频部分即作为中低分辨率的DEM,并对该方法及常用的重采样方法进行对比分析。 (3)提出了一种基于多进制小波与插值结合的DEM尺度下推算法。首先利用多进制小波分解,将得到的高频部分进行双线性插值并与原始DEM数据做为低频的部分,通过多进制小波逆变换得到尺度下推后的DEM数据,并对实验结果进行了主客观评价。 (4)提出了一种多进制小波与滤波的DEM尺度下推算法。顾及多进制小波的方向性,将方向滤波作用于DEM数据上,构建DEM高频部分,并与原始DEM数据通过多进制小波重构得到尺度下推的DEM数据,并与上一算法进行比较分析。 (5)在对基于学习的图像超分辨率重建算法分析的基础上,提出了一种基于DEM非局部相似性约束的邻域重构DEM尺度下推算法。通过获取部分高分辨的DEM数据,基于子区域的相似性和领域相容性,在低分辨率DEM数据中,寻找待尺度下推数据的非局部相似子区域,根据相似程度,将高分辨率数据映射到对应的区域,从而完成DEM的尺度下推,并与插值算法及多进制小波重构的DEM尺度下推算法进行对比分析。
[Abstract]:As the basic data of national geographic information, Digital elevation Model (demm) is the frame data of National Spatial data Infrastructure (NSDI), which has important applications in national economy and national defense construction. Dem scale problem in the coexistence of DEM with different precision is a hot issue that needs to be solved urgently, which is not only the fundamental guarantee of integration of DEM and geoscience model. It is also the key of DEM data popularization and application, so Dem resolution calculation and conversion become one of the important research propositions of DEM producers and application departments. Starting from the original scattered point data, this paper explores a resolution calculation method that meets the requirements of DEM topographic representation. When the calculated resolution does not match with the actual application, scale conversion is often required. In view of this problem, this paper is based on fractal. Wavelet and other mathematical theories, combined with the terrain representation and terrain analysis of DEM, systematically study the resolution calculation and scale conversion of DEM. The main research results are as follows:. 1) this paper puts forward the DEM resolution determination method based on fractal theory. Dem is the representation of surface morphology, which should reflect the maximum amount of topographic information, which needs to start with the original data. The determination method of DEM resolution is explored. The relationship between DEM resolution and fractal information dimension is established by using fractal quantification to express the characteristics of terrain self-similarity and complexity. The inflection point which can describe the topographic information to the maximum extent is found by the linear slope difference, and the horizontal resolution of DEM is determined. In this paper, the method of DEM scaling estimation based on multiary wavelet decomposition is studied. Based on the practical application requirements of DEM, the basic principle of DEM scale upscaling is established. The factors of DEM scale push-up accuracy are taken into account. In this paper, a DEM scale upscaling method based on random number is constructed. By using the multi-resolution analysis and multi-scale analysis characteristic of the multi-ary wavelet, a DEM scale up-scaling method based on the multi-ary wavelet decomposition is proposed, which is based on the multi-ary wavelet decomposition of DEM. The obtained low-frequency part is regarded as the low resolution DEM, and the method and the resampling method are compared and analyzed. In this paper, we propose a DEM scaling algorithm based on the combination of multiary wavelet and interpolation. Firstly, the bilinear interpolation of the obtained high frequency part and the original DEM data are used as the low-frequency part by using the multi-ary wavelet decomposition. The DEM data of the scale are obtained by inverse wavelet transform, and the experimental results are evaluated objectively and subjectively. In this paper, a DEM scale estimation method based on multiary wavelet and filter is proposed. Considering the directivity of multiary wavelet, the directional filtering is applied to the DEM data, and the high frequency part of DEM is constructed. The scale derived DEM data are reconstructed from the original DEM data by multi-ary wavelet, and compared with the previous algorithm. 5) based on the analysis of Learn-based super-resolution image reconstruction algorithm, a neighborhood reconstruction DEM scale extrapolation method based on DEM nonlocal similarity constraints is proposed. Some high-resolution DEM data are obtained. Based on the similarity of subregions and domain compatibility, the non-local similar sub-regions of the data to be pushed down to scale are found in low-resolution DEM data, and the high-resolution data are mapped to the corresponding regions according to the similarity degree. Thus, the scale deduction of DEM is completed, and compared with the interpolation algorithm and the DEM scale extrapolation method of multi-ary wavelet reconstruction.
【学位授予单位】：南京师范大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：P208

【参考文献】