首页> 外文学位 >Wavelet representation of geodetic operators.
【24h】

Wavelet representation of geodetic operators.

机译:大地测量算子的小波表示。

获取原文
获取原文并翻译 | 示例

摘要

The main objective of this research is to introduce an alternative to the FFT computational scheme using the wavelet transform for the numerical evaluation of different geodetic operators. The new wavelet representation is built on orthogonal wavelet base functions. Eight geodetic operators are evaluated in this thesis: they are classified into direct geodetic integrals, inverse geodetic integrals, and the inversion of integrals. The direct geodetic integrals are the Stokes, the Vening Meinesz, the Poisson (upward continuation), and the terrain correction integrals. The inverse geodetic integrals are the inverse Vening Meinesz integral and the deflection-geoid formula. The Stokes and Poisson (downward continuation) integrals are inverted in the wavelet domain by a conjugate gradient method.;In each case, the role of the kernel's singularity in the wavelet multi-resolution analysis is studied. The integrals are approximated in finite multi-resolution analysis subspaces. A new implementation is introduced to decrease the computational effort. The full solution with all equations requires a large computer memory. Multi-resolution properties of the wavelet transform are used to divide the full solution into parts. Each part represents a level of wavelet detail coefficients or the approximation coefficients. Hard thresholding is used for the compression of the kernels' wavelet detail coefficients. Global fixed thresholding and level/direction-wise thresholding is tested for different kernels. High compression levels are achieved with an acceptable accuracy, which leads to large savings in computer memory and storage space required for allocating the matrices, and also the ability to work with sparse matrices. In the case of the inversion of the integrals, a set of equations is formed and solved using an iterative gradient method. Soft thresholding is used for de-noising stationary and non-stationary noise because of its smoothing properties. Conclusions and recommendations are given with respect to the suitability, accuracy, and efficiency of these methods.
机译:这项研究的主要目的是介绍一种使用小波变换的FFT计算方案的替代方法,用于对不同大地测量算符进行数值评估。新的小波表示基于正交小波基函数。本文对八个大地测量算子进行了评估:将它们分为直接大地积分,反大地积分和积分反演。大地测量的直接积分是斯托克斯,维涅斯(Meinesz),泊松(向上延伸)和地形校正积分。大地逆积分是维涅斯逆积分和偏大地线公式。通过共轭梯度法在小波域中反转Stokes和Poisson(向下连续)积分。在每种情况下,都研究了核奇异性在小波多分辨率分析中的作用。积分在有限的多分辨率分析子空间中近似。引入了新的实现方式以减少计算量。所有方程式的完整解决方案需要大的计算机内存。小波变换的多分辨率属性用于将完整解决方案分为多个部分。每个部分代表小波细节系数或近似系数的级别。硬阈值用于压缩内核的小波细节系数。针对不同的内核测试了全局固定阈值和级别/方向阈值。以可接受的精度获得高压缩级别,这将导致节省大量计算机内存和分配矩阵所需的存储空间,以及处理稀疏矩阵的能力。在积分求逆的情况下,使用迭代梯度法形成并求解一组方程。软阈值由于其平滑特性而用于消除固定噪声和非固定噪声。针对这些方法的适用性,准确性和效率给出了结论和建议。

著录项

  • 作者

    Elhabiby, Mohamed Mamdouh.;

  • 作者单位

    University of Calgary (Canada).;

  • 授予单位 University of Calgary (Canada).;
  • 学科 Geodesy.
  • 学位 Ph.D.
  • 年度 2007
  • 页码 148 p.
  • 总页数 148
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号