Scholarlink

<p>Recent advancements in Multiple-Point Statistics (MPS) simulations have focused on achieving realism by improving spatial continuity, conditioning data integration, stochasticity, and computational efficiency. However, simultaneously optimizing these factors remains a challenge. This work proposes a novel approach that integrates the strengths of Cross-Correlation Simulation (CCSIM) and the Discrete Wavelet Transform (DWT) to address these limitations. The method leverages the DWT's ability to capture essential spatial patterns in a compressed domain. It computes the DWT for both the training image (TI) and a region shared with previously simulated grids at a specific decomposition level. Similarity between these compressed representations is efficiently measured using a Cross-Correlation Function (CCF). By utilizing the compressed wavelet coefficients, the proposed method accelerates the search for the best-matching pattern while retaining the essential characteristics of the original data. Once identified, the original pattern is reconstructed by integrating the DWT detail coefficients through an Inverse-DWT transformation. Experiments conducted on diverse categorical training images demonstrate that the proposed approach achieves performance comparable to multi-scale CCSIM (MS-CCSIM) while exhibiting improvements in spatial continuity and computational efficiency. The open-source implementation of the method is available at <a href="https://github.com/MBS1984/CCWSIM">https://github.com/MBS1984/CCWSIM</a> for further development.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

CCWSIM: an efficient and fast wavelet-based CCSIM for categorical characterization of large-scale geological domains

  • 期刊论文

  • 发表日期：

    2025年4月2日

查看全文
  • Mojtaba Bavandsavadkoohi,
  • Erwan Gloaguen,
  • Behzad Tokhmechi,
  • Alireza Arab-Amiri,
  • Bernard Giroux

摘要

Recent advancements in Multiple-Point Statistics (MPS) simulations have focused on achieving realism by improving spatial continuity, conditioning data integration, stochasticity, and computational efficiency. However, simultaneously optimizing these factors remains a challenge. This work proposes a novel approach that integrates the strengths of Cross-Correlation Simulation (CCSIM) and the Discrete Wavelet Transform (DWT) to address these limitations. The method leverages the DWT's ability to capture essential spatial patterns in a compressed domain. It computes the DWT for both the training image (TI) and a region shared with previously simulated grids at a specific decomposition level. Similarity between these compressed representations is efficiently measured using a Cross-Correlation Function (CCF). By utilizing the compressed wavelet coefficients, the proposed method accelerates the search for the best-matching pattern while retaining the essential characteristics of the original data. Once identified, the original pattern is reconstructed by integrating the DWT detail coefficients through an Inverse-DWT transformation. Experiments conducted on diverse categorical training images demonstrate that the proposed approach achieves performance comparable to multi-scale CCSIM (MS-CCSIM) while exhibiting improvements in spatial continuity and computational efficiency. The open-source implementation of the method is available at https://github.com/MBS1984/CCWSIM for further development.