基于岩石高光谱图像迁移学习的岩性智能识别

doi:10.3964/j.issn.1000-0593(2025)08-2289-13

摘要
参考文献
相关文章 (15)

全文: PDF (36767 KB)
输出: BibTeX | EndNote (RIS)

摘要：岩性快速识别具有重要的基础地质研究意义与工程应用价值。传统岩性识别主要依赖于岩石图像特征，但在识别外观相似的岩石时易出现混淆。为此，相关研究进一步利用光谱特征以反映岩石组成信息，但光谱测试通常需要制样，且属于有损测试。为此，凭借成像高光谱技术图谱合一、无损非接触的优势，提出了一种基于岩石高光谱图像迁移学习的岩性智能识别方法。首先，对岩石感兴趣区域的高光谱数据进行归一化和降维处理，降低光谱数据的冗余度。然后，结合三维残差网络（3D-ResNet）构建岩石高光谱图像迁移学习模型，利用残差网络提取三维信息。再利用迁移学习的方法，通过加载预训练权重对模型进行训练，从而实现岩性智能识别。采用混淆矩阵、准确率（ACC）、精确率（P）、召回率（R）、F₁值（F₁）作为模型准确率的评价指标，针对基于ResNet-101和ResNet-18/34/50的模型开展对比分析。结果表明，ResNet-101迁移模型在测试集的准确率最高，可达98.29%。P最高可达98.32%，R最高可达98.29%，F₁最高可达98.31%。ResNet-101对岩石光谱数据识别的准确率基本达到90%以上（除了绿泥石片岩），大部分结果甚至可以达到100%。相较于ResNet-18/34/50，ResNet-101识别精度更高且对每类岩石识别的稳定性更好。此外，利用本方法对隧道现场采样岩石进行了逐个像素点的岩性预测，验证了本文提出的岩性智能识别方法具有良好的鲁棒性和泛化性，可用于地质、测井、交通、水利等工程中的岩性快速智能识别。

关键词：高光谱成像；岩性识别；迁移学习；三维残差神经网络；光谱特征

Abstract：The rapid identification of lithology holds significant fundamental geological research significance as well as engineering application value. Traditional lithology recognition primarily depends on the image features of rocks. However, confusion tends to arise when identifying rocks with similar appearances. Consequently, relevant studies further utilize spectral features to reflect the compositional information of rocks. Nevertheless, spectral testing usually demands sample preparation and belongs to the category of destructive testing. This article proposes an intelligent lithology recognition method based on transfer learning of rock hyperspectral images, taking advantage of the integrated imaging hyperspectral technology and the non-destructive, non-contact imaging characteristics. Firstly, the hyperspectral data of the rock region of interest are normalized, and dimensionality reduction is performed to reduce the redundancy of spectral data. Then, a rock hyperspectral image transfer learning model is constructed using a 3D ResNet network, and three-dimensional information is extracted through a residual network. The transfer learning method is reused to train the model by loading pre-trained weights, thereby achieving intelligent recognition of lithology. In this article, the confusion matrix, accuracy (ACC), precision (P), recall (R), and F1 values (F1) are used as evaluation indicators for model accuracy. A comparative analysis is conducted on ResNet101 and ResNet18/34/50 models. The results indicate that the ResNet-101 migration model has the highest accuracy in the test set, reaching 98.29%. The highest P can reach 98.32%, the highest R can reach 98.29%, and the highest F1 can reach 98.31%. The accuracy of ResNet-101 in identifying rock spectral data is over 90% (except for chlorite schist), and most results can even reach 100%. Compared to ResNet18/34/50, ResNet101 has higher recognition accuracy and better stability for identifying each type of rock. In addition, this method was employed to predict the lithology of sampled tunnel site rocks pixel by pixel, verifying the good robustness and generalization performance of the proposed lithology intelligent identification method, which can be used for rapid and intelligent lithology identification in engineering fields like geology, logging, transportation, and water conservancy.

Key words：Hyperspectral imaging; Rock identification; Transfer learning; 3D residual neural network; Spectral characteristics

收稿日期: 2024-11-11 修订日期: 2025-03-24

中图分类号:

P588

基金资助: 国家自然科学基金优秀青年科学基金项目(52022053), 国家自然科学基金面上项目（52379103, 52279103）资助

通讯作者: 许振浩 E-mail: zhenhao_xu@sdu.edu.cn

作者简介: 李珊，女，1999年生，山东大学齐鲁交通学院博士研究生 e-mail: 202135410@mail.sdu.edu.cn

引用本文:

李珊，林鹏，许振浩，向航，李千纪. 基于岩石高光谱图像迁移学习的岩性智能识别[J]. 光谱学与光谱分析, 2025, 45(08): 2289-2301.
LI Shan, LIN Peng, XU Zhen-hao, XIANG Hang, LI Qian-ji. Intelligent Lithology Identification Based on Transfer Learning of Rock Hyperspectral Images. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(08): 2289-2301.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)08-2289-13 或 https://www.gpxygpfx.com/CN/Y2025/V45/I08/2289

[1] Shi H, Xu Z H, Lin P, et al. Geoenergy Science and Engineering, 2023, 231: 212382.
[2] Liang H B, Chen H F, Guo J H, et al. Expert Systems with Applications, 2022, 189: 116142.
[3] Song L, Yin X Y, Yin L J. IEEE Geoscience and Remote Sensing Letters, 2023, 20: 7503105.
[4] XU Zhen-hao, MA Wen, LI Shu-cai, et al（许振浩，马文，李术才，等）. Geological Review（地质论评）, 2022, 68(6): 2290.
[5] Xu Z H, Ma W, Lin P, et al. Journal of Rock Mechanics and Geotechnical Engineering, 2022, 14(4): 1140.
[6] LI Juan, SUN Hui-lan, HOU Qing-xiang（李娟，孙惠兰，侯庆香）. China Petrochem（中国石油石化）, 2017, 10: 61.
[7] Hou S K, Shi H Y, Cao X H, et al. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 5521213.
[8] Farmonov N, Amankulova K, Szatmári J, et al. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2023, 16: 1576.
[9] ZHONG Jia-ping, LI Yun-song, XIE Wei-ying, et al（钟佳平，李云松，谢卫莹，等）. Acta Electronica Sinica（电子学报）, 2024, 52(5): 1716.
[10] YAN Shou-xun, ZHANG Bing, ZHAO Yong-chao, et al(燕守勋, 张兵, 赵永超, 等). Remote Sensing Technology and Applications（遥感技术与应用）, 2003, 18(4): 191.
[11] WU Meng-juan, JIN Jia, WANG Jin-lin, et al(吴梦娟，靳佳，王金林，等). Acta Geologica Sinica(地质学报)，2024，98(1)：314.
[12] WANG Jun-jie, YUAN Xi-ping, GAN Shu, et al(王俊杰, 袁希平, 甘淑, 等). Journal of Lanzhou University (Natural Science Edition)[兰州大学学报(自然科学版)], 2023, 59(6): 786.
[13] Lin N, Fu J W, Jiang R Z, et al. Remote Sensing, 2023, 15(15): 3764.
[14] LI Lian-jie, FAN Shu-xiang, WANG Xue-wen, et al(李廉洁, 樊书祥, 王学文, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2022, 42(4): 1250.
[15] Xu Y H, Wang H, Zhou F, et al. IEEE Transactions on Geoscience and Remote Sensing, 2025, 63: 5511216.
[16] He L, Li J, Liu C Y, et al. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(3): 1579.
[17] Li Z Y, Xue Z H, Xu Q, et al. IEEE Transactions on Geoscience and Remote Sensing, 2024, 62: 5502019.
[18] ZHAO Xin, MA Jing-yi, CHEN Han, et al（赵昕，马竞一，陈晗，等）. Transactions of the Chinese Society for Agricultural Machinery（农业机械学报）, 2024, 55(4): 368.
[19] ZHOU Fei-xiang, JIANG Hong, GUO Bao-lin, et al（周飞翔，姜红，郭宝林，等）. China Journal of Chinese Materia Medica（中国中药杂志）, 2024, 49(24): 6660.
[20] Okada N, Maekawa Y, Owada N, et al. Minerals, 2020, 10(9): 809.
[21] Afjal M I, Mondal M N I, Mamun M A. Journal of Spatial Science, 2024, 69(3): 821.
[22] Mei S H, Ji J Y, Hou J H, et al. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(8): 4520.
[23] Ma X T, Man Q X, Yang X M, et al. Remote Sensing, 2023, 15(4): 992.
[24] LIU Ye, HAN Yu-bo, ZHU Wen-rui（刘烨，韩雨伯，朱文瑞）. Earth Science Fronties（地学前缘）, 2024, 31(4): 95.
[25] Liu L F, Ji M, Buchroithner M. Sensors, 2018, 18(9): 3169.
[26] Galdames F J, Perez C A, Estevez P A, et al. Chemometrics and Intelligent Laboratory Systems, 2022, 224: 104538.
[27] Yang K, Zhao M, Argyropoulos D. Postharvest Biology and Technology, 2025, 219: 113247.
[28] Xu Z H, Shi H, Lin P, et al. International Journal of Rock Mechanics and Mining Sciences, 2024, 180: 105814.
[29] Xu Z H, Yu T F, Lin P, et al. Engineering Geology, 2023, 325: 107279.