With the deepening of energy resource exploration in China, traditional lithology identification methods have become inadequate for the efficient and accurate analysis of massive core samples. The development of artificial intelligence technology has provided new insights for intelligent lithology identification. However, it still faces challenges under complex geological conditions, such as blurred lithological boundaries and insufficient multi-scale feature fusion.To address these issues, an innovative method combining deep learning with fine-grained labeling technology was proposed. Residual fully connected networks (both serial and parallel) were innovatively constructed, and the vanishing gradient problem in deep networks was effectively solved by these networks. Group-point and odd-even row-column sampling strategies were further designed. Through these strategies, the traditional point-to-point prediction mode was broken through, and the model’s receptive field for regional lithological features was significantly expanded. In addition, a secondary fine-grained labeling system based on K-means clustering and expert knowledge was established, and the precise capture of complex geological features such as transition zones and fractures was enabled. Experimental results show that using multi-dimensional RGB (red-green-blue)image features and a parallel network structure with a 19×19 sampling window, the model achieves an accuracy of 80% and a Macro-F1 score of 72.5% on the test set. This represents a significant improvement over baseline models. Notably, the recognition accuracy for monzogranitic gneiss exceeds 85%. This method maintains the model’s lightweight nature while improving training efficiency by approximately 47%, demonstrating broad application prospects. It can be scaled for the automated interpretation of archived core samples to significantly reduce labor costs. Looking forward, it provides core technical support for deploying edge computing devices directly at exploration sites, driving geological exploration toward a real-time, intelligent paradigm and opening a new path for the deep integration of geoinformatics and artificial intelligence.
ZHANG Yan, ZHANG Ling-yuan, ZHANG Hong-bing, SUN Jiang-bing, TANG Ze-yu, HE Zi-juan, LUO Cong.
2D Lithology Intelligent Recognition Method of Core Images in Library Collections Based on Lightweight Network[J].
Science Technology and Engineering, 2026, 26(30): 12926-12941 DOI:10.12404/j.issn.1671-1815.2507503
XuZhenhao, MaWen, LiShucai, et al. Lithology identification: methods, current situation and intelligent development trends[J]. Geological Review, 2022, 68(6): 2290-2304.
[5]
LarsonT E, SivilJ E, PeriwalP, et al. A machine-learning workflow to integrate high-resolution core-based facies into basin-scale stratigraphic models for the wolfcamp and third bone spring sand, delaware basin[J]. Journal of Subsurface Characterization, 2023, 11(4). DOI: 10.1190/INT-2023-0009.1.
[6]
ZhangX, ZhangH, ChangF, et al. Application of corrected met-hods for high-resolution XRF core scanning elements in lake sediments[J]. Applied Sciences, 2020, 10(22). DOI: 10.3390/app10228012.
[7]
KhodadadzadehM, GloaguenR. Upscaling high-resolution mineralogical analyses to estimate mineral abundances in drill core hyperspectral data[C]// IEEE International Geoscience and Remote Sensing Symposium. Yokohama: IEEE, 2019: 1845-1848.
GuanYao, WangQinghui, FengJin, et al. Integrated lithological identification of altered igneous rocks based on machine learning: a case study of the Huizhou 26-6 well area in the Pearl River Mouth Basin of the South China Sea[J]. Journal of Jilin University (Earth Science Edition), 2024, 54(1): 345-358.
SunQian, ZhaoJunlong, ZhangYuchen, et al. Rock type identification of tight sandstone reservoirs based on the SSMO-SSA-LGBM algorithm[J]. Geological Bulletin of China, 2025, 44(5): 935-948.
ZengRui, ShaoYanlin, HuangYu, et al. Intelligent identification of logging lithology based on deep learning[J]. Science Technology and Engineering, 2025, 25(30): 12804-12812.
[16]
XuZ, MaW, LinP, et al. Deep learning of rock images for intelligent lithology identification[J]. Computers & Geosciences, 2021, 154: 104799.
[17]
XuZ, ShiH, LinP, et al. Integrated lithology identification based on images and elemental data from rocks[J]. Journal of Petroleum Science & Engineering, 2021, 205: 108853.
[18]
XuZ, ShiH, LinP, et al. Intelligent on-site lithology identification based on deep learning of rock images and elemental data[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19: 1-5.
TanYongjian, TianMiao, XuDexin, et al. Research on rock image classification and recognition based on xception network[J]. Geography and Geo-Information Science, 2022, 38(3): 17-22.
[21]
RanX, XueL, ZhangY, et al. Rock classification from field image patches analyzed using a deep convolutional neural network[J]. Mathematics, 2019, 7(8). DOI: 10.3390/math7080755.
[22]
LiuX, WangH, JingH, et al. Research on intelligent identification of rock types based on faster R-CNN method[J]. IEEE Access, 2020, 8: 21804-21812.
[23]
AlzubaidiF, MostaghimiP, SwietojanskiP, et al. Automated lithology classification from drill core images using convolutional neural networks[J]. Journal of Petroleum Science & Engineering, 2021, 197. DOI: 10.1016/j.petrol. 2020.107933.
XiongYuehan, LiuDongyan, LiuDongsheng, et al. An automatic rock classification method based on deep learning of microscopic images of rock samples[J]. Journal of Jilin University (Earth Science Edition), 2021, 51(5): 1597-1604.
[26]
YanB, ZhengL, WangX, et al. Deep learning-based recognition method of red bed soft rock image[J]. Geological Journal, 2023, 58(6): 2418-2426.
YanYanfang, ShaoYanlin, WangQing, et al. Lithology identification of clastic rock in UAV images based on data fusion[J]. Science Technology and Engineering, 2024, 24(12): 4869-4875.
[29]
HouZ, WeiJ, ShenJ, et al. Intelligent lithology identification methods for rock images based on object detection[J]. Natural Resources Research, 2023, 32(6): 2965-2980.
[30]
RodgerA, FabrisA, LaukampC. Feature extraction and clustering of hyperspectral drill core measurements to assess potential lithological and alteration boundaries[J]. Minerals, 2021, 11(2). DOI: 10.3390/min11020136.
HaoHuizhen, GuQing, HuXiumian. Research progress and prospects of intelligent mineral recognition methods based on machine learning[J]. Earth Science, 2021, 46(9): 3091-3106.
[33]
MüllerS, MeimaJ A. Mineral classification of lithium-bearing pegmatites based on laser-induced breakdown spectroscopy: application of semi-supervised learning to detect known minerals and unknown material[J]. Spectrochimica Acta, Part B: Atomic Spectroscopy, 2022, 189. DOI: 10.1016/j.sab. 2022.106370.
YangBiao, NiRuipu, GaoHao, et al. Automatic extraction of mineral features based on multi-resolution images and intelligent recognition model for minerals[J]. Nonferrous Metals, 2022, 12(5): 84-93.
LiChunsheng, LiuTao, LiuZongbao, et al. Image identification of dense reservoir rock thin sections based on the spatial attention mechanism of Mask R-CNN[J]. Journal of China University of Petroleum (Natural Science Edition), 2024, 48(4): 24-32.
RenShujie, HuYong, HeWenxiang, et al. Microscopic image recognition of sandstone components based on deep learning[J]. Science Technology and Engineering, 2024, 24(9): 3727-3736.
LiuYe, HanYubo, ZhuWenrui. Identification and intelligent interpretation of mineral components, as well as information sharing and transfer learning among different rock types[J]. Earth Science Frontiers, 2024, 31(4): 95-111.