Tunnel and Underground Space (터널과지하공간)

Korean Society for Rock Mechanics and Rock Engineering (한국암반공학회)
권호 표지
DOI QR코드

DOI QR Code

Pixel-level Crack Detection in X-ray Computed Tomography Image of Granite using Deep Learning

딥러닝을 이용한 화강암 X-ray CT 영상에서의 균열 검출에 관한 연구

  • Hyun, Seokhwan (School of Civil and Environmental Engineering, Yonsei University) ;
  • Lee, Jun Sung (School of Civil and Environmental Engineering, Yonsei University) ;
  • Jeon, Seonghwan (School of Civil and Environmental Engineering, Yonsei University) ;
  • Kim, Yejin (School of Civil and Environmental Engineering, Yonsei University) ;
  • Kim, Kwang Yeom (Korea Institute of Civil Engineering and Building Technology) ;
  • Yun, Tae Sup (School of Civil and Environmental Engineering, Yonsei University)
  • 현석환 (연세대학교 건설환경공학과) ;
  • 이준성 (연세대학교 건설환경공학과) ;
  • 전성환 (연세대학교 건설환경공학과) ;
  • 김예진 (연세대학교 건설환경공학과) ;
  • 김광염 (건설기술연구원 극한환경연구센터) ;
  • 윤태섭 (연세대학교 건설환경공학과)
  • Received : 2019.06.05
  • Accepted : 2019.06.24
  • Published : 2019.06.30
Download PDF
⟨ Previous Next ⟩

Abstract

This study aims to extract a 3D image of micro-cracks generated by hydraulic fracturing tests, using the deep learning method and X-ray computed tomography images. The pixel-level cracks are difficult to be detected via conventional image processing methods, such as global thresholding, canny edge detection, and the region growing method. Thus, the convolutional neural network-based encoder-decoder network is adapted to extract and analyze the micro-crack quantitatively. The number of training data can be acquired by dividing, rotating, and flipping images and the optimum combination for the image augmentation method is verified. Application of the optimal image augmentation method shows enhanced performance for not only the validation dataset but also the test dataset. In addition, the influence of the original number of training data to the performance of the deep learning-based neural network is confirmed, and it leads to succeed the pixel-level crack detection.

본 연구에서는 화강암 시편에서 수압 파쇄법에 의해 생성된 미세균열의 3차원 형상을 X-ray CT 영상과 딥러닝을 이용하여 추출하였다. 실험으로 생성된 미세균열은 X-ray CT 영상 상에서 일반적인 영상처리방법으로는 추출하기 매우 어렵고 육안으로만 관찰이 가능한 형태를 지닌다. 하지만 본 연구에서 제안한 합성곱 신경망(Convolutional neural network) 기반 인코더-디코더(Encoder-Decoder) 구조의 딥러닝 모델을 통해 미세균열을 정량적으로 추출할 수 있었다. 특히 픽셀 단위의 미세균열 추출을 위해 인코딩 과정에서 소실되는 정보를 디코딩 과정으로 직접 전달하는 디코더 모델을 제안하였다. 또한, 딥러닝 기반 신경망 학습에 필요한 데이터의 수를 증가시키기 위해 이미지의 분할(Division), 회전(Rotation), 그리고 반전(Flipping) 등으로 데이터를 생성하는 영상 증대 방법을 적용하였으며 이때 최적의 조합을 확인하였다. 최적의 영상 학습 데이터 증대 방법을 적용하였을 때 검증 데이터뿐만 아니라 테스트 데이터에서의 성능 향상을 확인하였다. 학습 데이터의 원본 개수가 딥러닝 기반 신경망의 균열 추출 성능에 미치는 영향을 확인하고 딥러닝 기술을 사용하여 성공적으로 미세균열을 추출하였다.

Keywords

OBGHBQ_2019_v29n3_184_f0001.png 이미지

Fig. 1. Example of the cross-sectional X-ray CT image of granite including micro-crack

OBGHBQ_2019_v29n3_184_f0002.png 이미지

Fig. 2. Crack detection by conventional image processing method. (a) an original image and (b) a ground-truth image and (c) the extracted crack by region growing method and (d) the extracted crack by locally adaptive thresholding method

OBGHBQ_2019_v29n3_184_f0003.png 이미지

Fig. 3. Overall architecture of the convolutional neural network based deep learning network for extracting micro-crack in X-ray CT image

OBGHBQ_2019_v29n3_184_f0004.png 이미지

Fig. 4. Recall, precision, and f-measure from validation data (a) with image augmentation and (b) without image augmentation

OBGHBQ_2019_v29n3_184_f0005.png 이미지

Fig. 5. Recall, precision, and f-measure from test data with respect to the number of images for training and validation

OBGHBQ_2019_v29n3_184_f0006.png 이미지

Fig. 6. Examples of crack extraction results using CNN-based neural network (a) original image from test data and (b) ground-truth image and (c) extracted crack image with 30 training and validation images and (d) extracted crack image with 90 training and validation images

OBGHBQ_2019_v29n3_184_f0007.png 이미지

Fig. 7. 3-D crack surface visualization (a) specimen including the extracted crack surface which propagated from a borehole. Extracted crack surface from (b) the front and (c) the side

Table 1. Crack detection performance by data augmentation using image division

OBGHBQ_2019_v29n3_184_t0001.png 이미지

Table 2. Crack detection performance by data augmentation using image rotation on the original 30 images

OBGHBQ_2019_v29n3_184_t0002.png 이미지

Table 3. Crack detection performance by data augmentation using image division and rotation

OBGHBQ_2019_v29n3_184_t0003.png 이미지

Table 4. Crack detection performance by data augmentation using image flipping

OBGHBQ_2019_v29n3_184_t0004.png 이미지

References

  1. Badrinarayanan, V., Kendall, A., and Cipolla, R., 2017, Segnet: A deep convolutional encoder-decoder architecture for image segmentation. IEEE transactions on pattern analysis and machine intelligence, Vol. 39, No. 12, pp. 2481-2495. https://doi.org/10.1109/TPAMI.2016.2644615
  2. Bandyopadhyay, O., Biswas, A., and Bhattacharya, B. B., 2016, Long-bone fracture detection in digital X-ray images based on digital-geometric techniques. Computer methods and programs in biomedicine, Vol. 123, pp. 2-14. https://doi.org/10.1016/j.cmpb.2015.09.013
  3. Bang, S., Park, S., Kim, H., and Kim, H., 2019, Encoder-decoder network for pixel‐level road crack detection in black‐box images. Computer‐Aided Civil and Infrastructure Engineering.
  4. Bennour, Z., Ishida, T., Nagaya, Y., Chen, Y., Nara, Y., Chen, Q., Sekine, K., and Nagano, Y., 2015, Crack extension in hydraulic fracturing of shale cores using viscous oil, water, and liquid carbon dioxide. Rock Mechanics and Rock Engineering, Vol. 48, No. 4, pp. 1463-1473. https://doi.org/10.1007/s00603-015-0774-2
  5. Cha, Y. J., Choi, W., Suh, G., Mahmoudkhani, S., and Buyukozturk, O., 2018, Autonomous structural visual inspection using region‐based deep learning for detecting multiple damage types. Computer‐Aided Civil and Infrastructure Engineering, Vol. 33, No. 9, pp. 731-747. https://doi.org/10.1111/mice.12334
  6. Goodman, R. E., 1989, Introduction to rock mechanics, Vol. 2, New York: Wiley.
  7. Ha, S. J., Choo, J., and Yun, T. S., 2018, Liquid CO 2 Fracturing: Effect of Fluid Permeation on the Breakdown Pressure and Cracking Behavior. Rock Mechanics and Rock Engineering, Vol. 51, No. 11, pp. 3407-3420. https://doi.org/10.1007/s00603-018-1542-x
  8. He, K., Zhang, X., Ren, S., and Sun, J., 2016, Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770-778.
  9. Iassonov, P., Gebrenegus, T., and Tuller, M., 2009, Segmentation of X‐ray computed tomography images of porous materials: A crucial step for characterization and quantitative analysis of pore structures. Water Resources Research, Vol. 45, No. 9.
  10. Ioffe, S., and Szegedy, C., 2015, Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv preprint arXiv:1502.03167.
  11. Jung, S., Zhuang, L., Yeom, S., Kim, K. Y., and Min, K., 2016, Influence of the Cleavage Anisotropy of Pocheon Granite on Hydraulic Fracturing Behaviour. Tunnel and Underground Space, Vol. 26, No. 4, pp. 327-337. https://doi.org/10.7474/TUS.2016.26.4.327
  12. Jeong, Y. J., Kim, K. Y., and Yun, T. S., 2017, A Study on Correlation between Heterogeneity Index and Mechanical Properties of Igneous Rocks using 3D X-ray Computed Tomography Image. Tunnel and Underground Space, Vol. 27, No. 5, pp. 333-342. https://doi.org/10.7474/TUS.2017.27.5.333
  13. Kang, T. H., Kim, K. Y., Park, D. W., and Shin, H. S., 2014, Influence of Anisotropy of Microcrack Distribution in Pocheon Granite Rock on Elastic Resonance Characteristics. The Journal of Engineering Geology, Vol. 24, No. 3, pp. 363-372. https://doi.org/10.9720/kseg.2014.3.363
  14. Ketcham, R. A., and Carlson, W. D., 2001, Acquisition, optimization and interpretation of X-ray computed tomographic imagery: applications to the geosciences. Computers and Geosciences, Vol. 27, No. 4, pp. 381-400. https://doi.org/10.1016/S0098-3004(00)00116-3
  15. Krizhevsky, A., Sutskever, I., & Hinton, G. E., 2012, Imagenet classification with deep convolutional neural networks. In Advances in neural information processing systems, pp. 1097-1105.
  16. Li, S., Zhao, X., and Zhou, G., 2019, Automatic pixel‐level multiple damage detection of concrete structure using fully convolutional network. Computer‐Aided Civil and Infrastructure Engineering.
  17. Liu, T., Zhang, X. N., Li, Z., and Chen, Z. Q., 2014, Research on the homogeneity of asphalt pavement quality using X-ray computed tomography (CT) and fractal theory. Construction and Building Materials, Vol. 68, pp. 587-598. https://doi.org/10.1016/j.conbuildmat.2014.06.046
  18. Long, J., Shelhamer, E., and Darrell, T., 2015, Fully convolutional networks for semantic segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 3431-3440.
  19. Noh, H., Hong, S., and Han, B., 2015, Learning deconvolution network for semantic segmentation. In Proceedings of the IEEE international conference on computer vision, pp. 1520-1528.
  20. Perez, L., and Wang, J., 2017, The effectiveness of data augmentation in image classification using deep learning. arXiv preprint arXiv:1712.04621.
  21. Schluter, S., Sheppard, A., Brown, K., and Wildenschild, D., 2014, Image processing of multiphase images obtained via X‐ray microtomography: a review. Water Resources Research, Vol. 50, No. 4, pp. 3615-3639. https://doi.org/10.1002/2014WR015256
  22. Simonyan, K., and Zisserman, A., 2014, Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.
  23. Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., and Rabinovich, A., 2015, Going deeper with convolutions. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 1-9.
  24. Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., and Wojna, Z., 2016, Rethinking the inception architecture for computer vision. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 2818-2826.
  25. Szegedy, C., Ioffe, S., Vanhoucke, V., and Alemi, A. A., 2017, Inception-v4, inception-resnet and the impact of residual connections on learning. In Thirty-First AAAI Conference on Artificial Intelligence.
  26. Yang, X., Li, H., Yu, Y., Luo, X., Huang, T., and Yang, X., 2018, Automatic pixel‐level crack detection and measurement using fully convolutional network. Computer‐Aided Civil and Infrastructure Engineering, Vol. 33, No. 12, pp. 1090-1109. https://doi.org/10.1111/mice.12412
  27. Zhang, A., Wang, K. C., Li, B., Yang, E., Dai, X., Peng, Y., Fei, Y., Liu, Y., Li, J. Q., and Chen, C., 2017, Automated pixel‐level pavement crack detection on 3D asphalt surfaces using a deep‐learning network. Computer‐Aided Civil and Infrastructure Engineering, Vol. 32, No. 10, pp. 805-819. https://doi.org/10.1111/mice.12297