铀矿地质2025，Vol.41Issue(6)：1097-1113,17.DOI:10.3969/j.issn.1000-0658.2025.41.076
岩心成像扫描与高光谱地质编录进展
Progress on Imaging Core Scan and Hyperspectral Geological Logging
潘蔚 1张川 1张元涛 2刘洪成 1刘鹏飞1
作者信息
- 1. 铀资源探采与核遥感全国重点实验室,北京 100029||核工业北京地质研究院,北京 100029
- 2. 核工业二八〇研究所,四川 广汉 618300
- 折叠
摘要
Abstract
The objectivity and integrity of core information directly affect its economic value and scientific significance.Traditional manual geological logging is professional and practical,but it is subjective and empirical.Therefore,more objective and intact technical methods have been explored at home and abroad.Digital imaging core scanning is currently recognized as an effective method.According to the working principle,it can be divided into three categories of true color three-dimensional photography,true color photography combined with spectral measurement and imaging spectral(hyperspectral)scanning.In this paper,we briefly summarized the advantages and disadvantages of the three types at first,and recognized that the hyperspectral scanning image can simultaneously study the core structure and material composition,therefore has the best application prospect.Then,we traced the development status of hyperspectral equipment and data analyzing in detail and found that two shortcoming hindered its wide engineering application,one is the bulky and expensive of the equipment which make its hard to meet the needs of economical and efficient drilling in engineering.The other is the stubborn application of alteration minerals mapping which helps the theoretical research on deposit genesis but fail to solve the key engineering problems such as geological logging.Subsequently,the authors analyzed the up to date progress of hyperspectral application since the thriving of deep learning,and believed that hyperspectral core scanning is expected to get the break through in automatic identification of core structure and lithology.Finally,some suggestions were proposed for the realization of automatic geologic logging with hyperspectral core scanning in three prospects.In manufacture,the equipment should be both vehicle-mounted for general system and portable for special field.In data processing and analyzing,infrared spectra of major minerals should be established together with the deep learning geoscience label library for the lithology identification.In automatic geological logging,multimodal deep learning algorithms should be developed to build the specific software systems based on domestically produced general-purpose large model.关键词
岩心/成像扫描/高光谱/地质编录/智能化Key words
core/imaging scanning/hyperspectral/geological logging/AI分类
地质学引用本文复制引用
潘蔚,张川,张元涛,刘洪成,刘鹏飞..岩心成像扫描与高光谱地质编录进展[J].铀矿地质,2025,41(6):1097-1113,17.
