基于钻孔电视图像的结构面调查是一种应用广泛的勘测方法, 可为岩体质量评价和工程设计提供基础数据支撑。 然而, 复杂地层中结构面形态各异, 宽度差异大且对比度低, 传统算法难以准确完整地识别。 为此, 该文提出一种基于改进的U-Net网络模型的钻孔电视图像结构面识别方法。 首先, 采用更深层次的编码-解码网络结构, 处理对比度低导致的结构面局部断裂问题, 并结合通道注意力机制和残差模块提高编码阶段各层级对结构面特征的提取能力。 其次, 在较低层级跳跃连接中引入多尺度空间注意力模块, 提高对复杂形态结构面处理能力, 丰富编码层结构面语义特征; 同时, 通道注意力也被用于充分融合来自编码层和解码层的多通道结构面信息。 然后, 通过利用与地质作用下地层变形类似的透视变形等方式扩增钻孔图像数据, 并结合焦点损失和Dice损失进行联合训练, 以减轻图像数据不平衡问题, 增强网络泛化能力。 最后, 将同一钻孔图像作为训练集, 邻近钻孔图像作为测试集进行消融和对比实验。 结果表明, 相比于已有相关网络模型, 该文所提方法能更有效地针对复杂地层钻孔图像, 较准确、 完整地识别各类岩体结构面, 精确率和召回率均超过77.00%, 比改进前的U-Net网络分割效果分别提高了7.96%和14.99%。 该文方法可为现场结构面钻孔调查以及岩体质量评价提供技术支撑。
Abstract
[Objective] Discontinuities are vital components of rock mass, significantly affecting its strength, deformation, and seepage characteristics. They provide essential parameters for rock mass classification and engineering design. Borehole television technology is a widely used method for capturing these discontinuities within the rock mass, offering high-resolution in situ images. However, in complex strata, the discontinuities appear in various morphologies with significant width differences. Coupled with the rapid texture changes on rock wall faces, these discontinuities create a highly uneven contrast, making it challenging for traditional algorithms to recognize them accurately. To address this challenge, this study introduces an improved deep learning network model specifically designed for borehole images of complex strata. [Methods] The proposed model, based on the U-Net architecture, incorporates a deeper encoding-decoding network structure. This structure effectively handles semantic information related to discontinuity breaks caused by uneven contrast. The model integrates channel attention mechanisms and residual modules, enhancing feature extraction capabilities at different levels in the encoding stage. In addition, the channel attention mechanism fuses multichannel discontinuity information from both encoding and decoding layers. A multiscale spatial attention module introduced in the lower-level skip connection improves the ability to process complex morphological discontinuities and enriches the semantic features of discontinuities in the coding layer. In this study, the borehole image data are augmented in various ways, such as using perspective deformation similar to the stratum deformation under geological action. This study also employs joint training with focal loss and Dice loss to handle imbalanced image data. The generalization ability of the network model is thoroughly validated through ablation studies and comparative experiments using the same borehole image as the training set and neighboring borehole images as the test set. For comprehensive quantitative evaluation, this study uses several metrics, including precision, recall, F1-Score, and F2-Score. [Results] Our experimental evaluation, conducted on a self-made borehole image dataset, indicated that compared to several common image segmentation network models, our proposed model significantly improved the recognition capability of rock discontinuities in borehole images from complex strata while ensuring faster computational efficiency. The precision and recall on the test set for the proposed model reached 78.23% and 77.85%, respectively. This marked an improvement in segmentation performance by 7.96% and 14.99%, respectively, compared with the basic U-Net model. Both the F1-Score and F2-Score were close to 78%. Although the model size was 18.13 MB and had approximately twice the parameters of the base U-Net, the deeper network hierarchy reduced the number of channels of shallow high-resolution feature maps, resulting in a reduction in computational load. The model achieved an FPS of 85, which was slightly higher than that of the basic U-Net model. [Conclusions] This study meticulously improves upon the basic U-Net model by strategically incorporating the attention mechanism, residual connections, and multiscale convolutions. The improved model exhibits high accuracy and robustness. It effectively confronts the challenges associated with balancing detailed features and high-level semantics owing to significant width differences in discontinuities within complex strata. Furthermore, it addresses issues related to incomplete extraction of discontinuities caused by uneven contrast between discontinuities and rock wall surfaces. As such, this improved model provides strong technical support for the automatic identification of rock discontinuities in on-site borehole investigations.
关键词
钻孔图像 /
结构面识别 /
深度学习 /
U-Net神经网络 /
注意力机制
Key words
borehole images /
rock discontinuity recognition /
deep learning /
U-Net network /
attention mechanism
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 吴顺川,李利平,张晓平.岩石力学[M].北京:高等教育出版社, 2021. WU S C, LI L P, ZHANG X P. Rock mechanics[M]. Beijing:Higher Education Press, 2021.(in Chinese)
[2] RIQUELME A J, ABELLÁN A, TOMÁS R, et al. A new approach for semi-automatic rock mass joints recognition from 3-D point clouds[J]. Computers&Geosciences, 2014, 68:38-52.
[3] ASSOUS S, ELKINGTON P, CLARK S, et al. Automated detection of planar geologic features in borehole images[J]. Geophysics, 2014, 79(1):D11-D19.
[4] 汪进超,王川婴,韩增强,等.基于钻孔摄像技术的珊瑚礁完整性评价分析[J].中南大学学报(自然科学版), 2016, 47(5):1619-1624. WANG J C, WANG C Y, HAN Z Q, et al. Analysis of coral reef integrity based on borehole camera technology[J]. Journal of Central South University (Science and Technology), 2016, 47(5):1619-1624.(in Chinese)
[5] BAE D S, KIM K S, KOH Y K, et al. Characterization of joint roughness in granite by applying the scan circle technique to images from a borehole televiewer[J]. Rock Mechanics and Rock Engineering, 2011, 44(4):497-504.
[6] 王川婴,钟声,孙卫春.基于数字钻孔图像的结构面连通性研究[J].岩石力学与工程学报, 2009, 28(12):2405-2410. WANG C Y, ZHONG S, SUN W C. Study of connectivity of discontinuities of borehole based on digital borehole images[J]. Chinese Journal of Rock Mechanics and Engineering, 2009, 28(12):2405-2410.(in Chinese)
[7] AL-SIT W, Al-NUAIMY W, MARELLI M, et al. Visual texture for automated characterisation of geological features in borehole televiewer imagery[J]. Journal of Applied Geophysics, 2015, 119:139-146.
[8] OTSU N. A threshold selection method from gray-level histograms[J]. IEEE Transactions on Systems, Man, and Cybernetics, 1979, 9(1):62-66.
[9] CANNY J. A computational approach to edge detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, PAMI-8(6):679-698.
[10] 吴剑,冯少孔,李宏阶.钻孔成像中结构面自动判读技术研究[J].岩土力学, 2011, 32(3):951-957. WU J, FENG S K, LI H J. Study of automatically extracting structural plane parameters from borehole images[J]. Rock and Soil Mechanics, 2011, 32(3):951-957.(in Chinese)
[11] LI L, YU C, HAN Z Q, et al. Automatic identification of the rock-soil interface and solution fissures from optical borehole images based on color features[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(10):3862-3873.
[12] 汪进超,王川婴,胡胜,等.孔壁钻孔图像的结构面参数提取方法研究[J].岩土力学, 2017, 38(10):3074-3080. WANG J C, WANG C Y, HU S, et al. A new method for extraction of parameters of structural surface in borehole images[J]. Rock and Soil Mechanics, 2017, 38(10):3074-3080.(in Chinese)
[13] 李清波,杜朋召.基于边缘阈值分割的钻孔图像RQD自动分析方法研究[J].岩土工程学报, 2020, 42(11):2153-2160. LI Q B, DU P Z. Automatic RQD analysis method based on information recognition of borehole images[J]. Chinese Journal of Geotechnical Engineering, 2020, 42(11):2153-2160.(in Chinese)
[14] 邹先坚,王川婴,韩增强,等.全景钻孔图像中结构面全自动识别方法研究[J].岩石力学与工程学报, 2017, 36(8):1910-1920. ZOU X J, WANG C Y, HAN Z Q, et al. Fully automatic identifying the structural planes with panoramic images of boreholes[J]. Chinese Journal of Rock Mechanics and Engineering, 2017, 36(8):1910-1920.(in Chinese)
[15] 夏丁,葛云峰,唐辉明,等.数字钻孔图像兴趣区域分割与岩体结构面特征识别[J].地球科学, 2020, 45(11):4207-4217. XIA D, GE Y F, TANG H M, et al. Segmentation of region of interest and identification of rock discontinuities in digital borehole images[J]. Earth Science, 2020, 45(11):4207-4217.(in Chinese)
[16] 张新钰,高洪波,赵建辉,等.基于深度学习的自动驾驶技术综述[J].清华大学学报(自然科学版), 2018, 58(4):438-444. ZHANG X Y, GAO H B, ZHAO J H, et al. Overview of deep learning intelligent driving methods[J]. Journal of Tsinghua University (Science and Technology), 2018, 58(4):438-444.(in Chinese)
[17] LI S T, SONG W W, FANG L Y, et al. Deep learning for hyperspectral image classification:An overview[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(9):6690-6709.
[18] HAO S J, ZHOU Y, GUO Y R. A brief survey on semantic segmentation with deep learning[J]. Neurocomputing, 2020, 406:302-321.
[19] ZHANG Y, CHEN J Q, LI Y L. Segmentation and quantitative analysis of geological fracture:A deep transfer learning approach based on borehole televiewer image[J]. Arabian Journal of Geosciences, 2022, 15(3):300.
[20] ZHANG Y, CHEN J Q, LI Y L, et al. Deep ensemble learning for quantitative geological fracture analysis using borehole televiewer images[J]. Journal of Applied Geophysics, 2023, 213:105046.
[21] HAN S Y, XIAO X, SONG B Y, et al. Automatic borehole fracture detection and characterization with tailored faster R-CNN and simplified Hough transform[J]. Engineering Applications of Artificial Intelligence, 2023, 126:107024.
[22] LIU C C, KIM J, SONG J J, et al. Intelligent recognition and identification of fracture types and parameters for borehole images based on developed convolutional neural networks and post-processing[J]. Engineering Fracture Mechanics, 2023, 292:109624.
[23] 黄达,钟助.基于单个钻孔孔壁电视图像确定地下岩体结构面产状的普适数学方法[J].地球科学--中国地质大学学报, 2015, 40(6):1101-1106. HUANG D, ZHONG Z. A universal mathematical method for determining the occurrence of underground rock discontinuity based on TV picture of wall of a single borehole rock discontinuity based on TV picture of wall of a single borehole[J]. Earth Science-Journal of China University of Geosciences, 2015, 40(6):1101-1106.(in Chinese)
[24] 韩增强,王川婴,周济芳,等.基于钻孔图像的孔壁岩体完整性计算及在裂隙岩体灌浆效果评价中的应用[J].岩土工程学报, 2016, 38(S2):245-249. HAN Z Q, WANG C Y, ZHOU J F, et al. Calculation of borehole wall rock integrity based on borehole images and its application in evaluation of grouting effect in fractured rock mass[J]. Chinese Journal of Geotechnical Engineering, 2016, 38(S2):245-249.(in Chinese)
[25] 张帅军,汪进超,韩增强.基于钻孔图像的结构面分布分维数计算方法[J].人民长江, 2020, 51(5):203-209. ZHANG S J, WANG J C, HAN Z Q. Fractal dimension calculation method on rock mass structural plane distribution based on borehole image[J]. Yangtze River, 2020, 51(5):203-209.(in Chinese)
[26] BADRINARAYANAN V, KENDALL A, CIPOLLA R. SegNet:A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12):2481-2495.
[27] RONNEBERGER O, FISCHER P, BROX T. U-Net:Convolutional networks for biomedical image segmentation[C]//18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Munich, Germany:Springer, 2015:234-241.
[28] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, USA:IEEE, 2018:7132-7141.
[29] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Las Vegas, USA:IEEE, 2016:770-778.
[30] LIU S T, HUANG D, WANG Y H. Receptive field block net for accurate and fast object detection[C]//Proceedings of the 15th European Conference on Computer Vision (ECCV). Munich, Germany:Springer, 2018:404-419.
[31] WOO S, PARK J, LEE J Y, et al. CBAM:Convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision. Munich, Germany:Springer, 2018:3-19.
[32] BLOICE M D, ROTH P M, HOLZINGER A. Biomedical image augmentation using Augmentor[J]. Bioinformatics, 2019, 35(21):4522-4524.
[33] 周迅,李永龙,周颖玥,等.基于改进DeepLabV3+网络的坝面裂缝检测方法[J].清华大学学报(自然科学版), 2023, 63(7):1153-1163. ZHOU X, LI Y L, ZHOU Y Y, et al. Dam surface crack detection method based on improved DeepLabV3+network[J]. Journal of Tsinghua University (Science and Technology), 2023, 63(7):1153-1163.(in Chinese)
[34] ZHOU Z W, SIDDIQUEE M M R, TAJBAKHSH N, et al. UNet++:A nested U-Net architecture for medical image segmentation[C]//4th International Workshop on Deep Learning in Medical Image Analysis and 8th International Workshop on Multimodal Learning for Clinical Decision Support. Granada, Spain:Springer, 2018:3-11.
[35] CHEN L C, PAPANDREOU G, SCHROFF F, et al. Rethinking atrous convolution for semantic image segmentation[J/OL]. arXiv.(2017-12-05)[2023-07-12]. arXiv:1706.05587v3.
[36] OKTAY O, SCHLEMPER J, LE FOLGOC L, et al. Attention U-Net:Learning where to look for the pancreas[J/OL]. arXiv.(2018-05-20)[2023-06-16]. arXiv:1804.03999v3.
[37] JHA D, SMEDSRUD P H, JOHANSEN D, et al. A comprehensive study on colorectal polyp segmentation with ResUNet++, conditional random field and test-time augmentation[J]. IEEE Journal of Biomedical and Health Informatics, 2021, 25(6):2029-2040.
基金
国家自然科学基金项目(51934003); 云南省重大科技专项(202102AF080001); 云南省创新团队资助项目(202105AE160023)
PDF(14787 KB)
