北京大学学报自然科学版 ›› 2023, Vol. 59 ›› Issue (2): 231-241.DOI: 10.13209/j.0479-8023.2022.108

上一篇    下一篇

基于生成对抗网络的砂岩薄片图像视野外重建

周嵘1,2, 吴朝东1,2,†, 张亚楠1,2   

  1. 1. 北京大学地球与空间科学学院, 北京 100871 2. 造山带与地壳演化教育部重点实验室, 北京 100871
  • 收稿日期:2022-03-31 修回日期:2022-07-04 出版日期:2023-03-20 发布日期:2023-03-20
  • 通讯作者: 吴朝东, E-mail: cdwu(at)pku.edu.cn
  • 基金资助:
    国家重大科技专项(2017ZX05008-001)资助

Outpainting Reconstruction of Sandstone Thin-Section Image Based on Generative Adversarial Network

ZHOU Rong1,2, WU Chaodong1,2,†, ZHANG Yanan1,2   

  1. 1. School of Earth and Space Sciences, Peking University, Beijing 100871 2. The Key Laboratory of Orogenic Belts and Crustal Evolution (MOE), Beijing 100871
  • Received:2022-03-31 Revised:2022-07-04 Online:2023-03-20 Published:2023-03-20
  • Contact: WU Chaodong, E-mail: cdwu(at)pku.edu.cn

PDF

摘要：

利用生成对抗网络(GAN)模型, 对砂岩薄片图像的微观颗粒和孔隙结构进行视野外重建, 并对预测图像的语义进行评价。结果表明, 模型能够预测2.25倍于原始视野的砂岩微观结构, 并且针对不同类型的岩石图像语义均具有良好的性能。模型对不同颗粒的表面纹理、颗粒形态以及多颗粒间复杂接触关系等语义的图像视野外预测结果与真实图像较为吻合。但是, 在微观特殊现象图像的视野外重建任务中, 模型缺乏对特殊现象的敏感性。在孔隙结构重建时, 模型对微孔面孔率的预测误差大于粒间孔、裂隙和溶蚀孔等孔隙空间。不同孔隙空间重建图像的预测效果可能与孔隙特征(如孔径大小和连通性)有关。

关键词: 生成对抗网络(GAN), 岩石薄片图像, 颗粒结构, 孔隙结构, 图像视野外推

Abstract:

A generative adversarial network (GAN) model is applied to the outpainting reconstruction on the micro grain and pore structure of sandstone thin-section image, and the semantics of the predicted image is analyzed in the model. The findings demonstrate that the model can predict a lager view of sandstone microstructure, which is 2.25 times the size of the original vision, and has good performance for different types of rock image semantics. The predicted image semantics, such as surface texture of different grains, grain morphology and complex contact relationship between multiple grains are consistent with the real results. However, in the task of outpainting reconstruction on microscopic special phenomenon in rock thin-section, the model lacks sensitivity to special phenomenon. In the task of reconstruction on pore structure, the prediction error of micropore distribution is larger than that of common pore spaces such as intergranular pores, fractures and dissolution pores. The prediction performance of reconstruction results of different pore spaces may be related to pore features (such as pore size and connectivity).

Key words: generative adversarial network (GAN), rock thin-section image, grain structure, pore structure, image outpainting

版权所有 © 《北京大学学报（自然科学版）》编辑部
技术支持：北京玛格泰克科技发展有限公司