摘要
岩石CT图像反映岩石孔隙和裂隙等结构特征,对CT图像进行处理并提取图像中的地质信息,对研究岩石的孔隙结构、分析岩石物理响应等具有重要意义.本文总结CT图像处理方法的原理、适用条件、应用效果及其影响因素,将图像处理算法分为基于数字图像的处理方法、基于机器学习的处理方法和超分辨率图像处理方法.结果表明:(1)数字图像处理方法简单易实现,应用范围广泛,但依赖于人工的先验知识;(2)基于机器学习的图像处理方法可自动、客观、准确、快速地提取图像信息,但岩石CT图像数据匮乏、模型泛化能力有待提升;(3)超分辨率图像处理方法能有效提升图像清晰度和细节,但受到岩样成像条件限制,分辨率提升效果有限.结合人工智能方法,为岩石CT图像的智能分析和自动识别带来新的可能性,加速地质信息的获取和解释过程.
CT images of rocks could reflect the structural characteristics of pores and fractures.Processing of CT images and extracting geological information are crucial for studying the pore structures of rocks and analyzing their physical responses,which would enhance our understanding of rock characteristics aiding in resource exploration and geophysical research.The principles,applicable conditions,application effects,and influencing factors of rock CT image processing methods were all reviewed.The image processing algorithms are categorized into three types,digital image-based processing methods,machine learning-based processing methods and super-resolution image processing methods.The results indicate that:(1)The digital image processing methods are straightforward and easy to implement,with a wide range of applications,however,rely heavily on prior knowledge from human operators;(2)Machine learning-based image processing methods could automatically,objectively,accurately,and quickly extract information from images,nevertheless,there is a shortage of rock CT image data and the generalization ability of models needs to be improved;(3)Super-resolution image processing methods could effectively improve the clarity and details of the image,but their effectiveness in enhancing resolution is limited by the imaging conditions of the rock samples.Integrating artificial intelligence methods offers the potential for the intelligent analysis and automatic recognition of rock CT images,which would accelerate the acquisition and interpretation of geological information.
作者
祝凰馨
张元中
ZHU HuangXing;ZHANG YuanZhong(State Key Laboratory of Petroleum Resources and Engineering,China University of Petroleum(Beijing),Beijing 102249,China)
出处
《地球物理学进展》
北大核心
2025年第5期1954-1976,共23页
Progress in Geophysics
基金
国家自然科学基金项目(41374144)
中石油“十四五”前瞻性基础性重大科技项目“测井数字岩石技术研究”(2021DJ4003)联合资助。
关键词
岩石CT图像
图像处理
深度学习
数字岩石物理
人工智能
Rock CT images
Image processing
Deep Learning
Digital rock physics
Artificial intelligence
