DOI QR코드

DOI QR Code

딥러닝과 영상처리기법을 이용한 콘크리트 지반 구조물 균열 탐지

Crack Detection of Concrete Structure Using Deep Learning and Image Processing Method in Geotechnical Engineering

  • 김아람 (한국건설기술연구원 인프라안전연구본부) ;
  • 김동현 (뉴로핏 주식회사) ;
  • 변요셉 (한국건설기술연구원 인프라안전연구본부) ;
  • 이성원 (한국건설기술연구원 인프라안전연구본부)
  • 투고 : 2018.12.14
  • 심사 : 2018.12.27
  • 발행 : 2018.12.31

초록

교량, 터널 옹벽 등의 콘크리트 구조물에서 수행되는 손상 조사 및 검사 방법은 일반적으로 검사원이 현장에서 직접 측량 도구를 사용하여 시각적으로 검사하는 방법이다. 이 방법은 검사원의 주관성에 크게 의존하기 때문에 기록의 객관성과 신뢰성이 떨어지게 된다. 따라서 균열을 자동으로 탐지하고 균열 특성을 객관적으로 분석할 수 있는 새로운 이미지분석기법이 필요하다. 본 연구에서는 콘크리트 이미지에서 균열을 검출하고 특성(균열의 길이, 폭)을 분석하기 위한 딥러닝 및 이미지분석기법을 개발하였다. 균열 검출과 해당 균열의 특성을 얻기 위해 두 가지 단계의 방법이 제안되었다. 제안된 방법의 성능을 검증하기 위하여 라벨이 있는 다양한 균열 이미지가 사용되었으며, 균열 판단과 구획화에 대해 90% 이상의 정확도를 확인하였다. 최종적으로 실제 촬영된 균열 영상의 균열 특성을 분석하고 실제 측정치와 오차를 확인하여 개발된 기법의 성능을 검증하였다.

The damage investigation and inspection methods performed in concrete facilities such as bridges, tunnels, retaining walls and so on, are usually visually examined by the inspector using the surveying tool in the field. These methods highly depend on the subjectivity of the inspector, which may reduce the objectivity and reliability of the record. Therefore, the new image processing techniques are necessary in order to automatically detect the cracks and objectively analyze the characteristics of cracks. In this study, deep learning and image processing technique were developed to detect cracks and analyze characteristics in images for concrete facilities. Two-stage image processing pipeline was proposed to obtain crack segmentation and its characteristics. The performance of the method was tested using various crack images with a label and the results showed over 90% of accuracy on crack classification and segmentation. Finally, the crack characteristics (length and thickness) of the crack image pictured from the field were analyzed, and the performance of the developed technique was verified by comparing the actual measured values and errors.

키워드

GJBGC4_2018_v34n12_145_f0001.png 이미지

Fig. 1. Image segmentation for crack detection

GJBGC4_2018_v34n12_145_f0002.png 이미지

Fig. 2. Database of concrete cracks (Özgenel, 2018)

GJBGC4_2018_v34n12_145_f0003.png 이미지

Fig. 3. Classification network for determination of crack presence

GJBGC4_2018_v34n12_145_f0004.png 이미지

Fig. 4. Segmentation network for concrete cracks

GJBGC4_2018_v34n12_145_f0005.png 이미지

Fig. 5. Application of thinning algorithm to crack image

GJBGC4_2018_v34n12_145_f0006.png 이미지

Fig. 6. Extending area for tracking crack data

GJBGC4_2018_v34n12_145_f0007.png 이미지

Fig. 7. Calculation method for profiling direction

GJBGC4_2018_v34n12_145_f0008.png 이미지

Fig. 8. Training loss and evaluation accuracy according to epoch for training and evaluation sets

GJBGC4_2018_v34n12_145_f0009.png 이미지

Fig. 9. Training loss and evaluation accuracy according to epoch for training and evaluation sets

GJBGC4_2018_v34n12_145_f0010.png 이미지

Fig. 10. Result of concrete crack segmentation

GJBGC4_2018_v34n12_145_f0011.png 이미지

Fig. 11. Result of segmentation, thinning, tracking and profiling

참고문헌

  1. Byun, T.B., Kim, J.H., and Kim, H.S. (2006), The Recognition of Crack Detection Using Difference Image Analysis Method based on Morphology, J. of the Korea Institute of Information and Communication Engineering, Vol.10, No.1, pp. 197-205.
  2. Cha, Y.J. and Choi, W. (2017), Vision-Based Concrete Crack Detection Using a Convolutional Neural Network, Dynamics of Civil Structures, Vol.2, pp.71-73.
  3. Cho, S., Kim, B., and Lee, Y.I. (2018), Image-Based Concrete Crack and Spalling Detection using Deep Learning, J. of the Korean Society of Civil Engineers, Vol.66, No.8, pp.92-97.
  4. Kim, J.W. and Jung, Y.W. (2017), Study on rapid structure visual inspection technology using drones and image analysis techniques for Damaged Concrete Structures, Proceeding of the Korean Society of Civil Engineers, pp.1788-1789.
  5. Kim, K.B. and Cho, J.H. (2010), Detection of Concrete Surface Cracks using Fuzzy Techniques, J. of the Korea Institute of Information and Communication Engineering, Vol.14, No.6, pp.1353-1358. https://doi.org/10.6109/jkiice.2010.14.6.1353
  6. Kim, Y. (2016), Development of Crack Recognition System for Concrete Structure Using Image Processing Method, J. of Korean Institute of Information Technology, Vol.14, No.10, pp.163-168.
  7. Lee, B.Y., Kim, Y.Y., and Kim, J.K. (2005), Development of Image Processing for Concrete Surface Cracks by Employing Enhanced Binarization and Shape Analysis Technique, J. of the Korea Concrete Institute, Vol.17, No.3, pp.361-368. https://doi.org/10.4334/JKCI.2005.17.3.361
  8. Lee, B.J., Shin, J.I., and Park, C.H. (2008), Development of Image Processing Program to Inspect Concrete Bridges, Proceedings of the Korea Concrete Institute, pp.189-192.
  9. Lee, J.H., Kim, I.H., and Jung, H.J. (2018), A Feasibility Study for Detection of Bridge Crack Based on UAV, Transactions of the Korean Society for Noise and Vibration Engineering, Vol.28, No.1, pp.110-117. https://doi.org/10.5050/KSNVE.2018.28.1.110
  10. Li W., Wang G., Fidon L., Ourselin S., Cardoso M.J., and Vercauteren T. (2017), On the Compactness, Efficiency, and Representation of 3D Convolutional Networks: Brain Parcellation as a Pretext Task. In: Niethammer M. et al. (eds) Information Processing in Medical Imaging. IPMI 2017. Lecture Notes in Computer Science, Vol. 10265. Springer, Cham.
  11. Ozgenel, C.F. (2018), "Concrete Crack Images for Classification", Mendeley Data, v1 http://dx.doi.org/10.17632/5y9wdsg2zt.1
  12. Park, H.S. (2013), Performance Analysis of the Tunnel Inspection System Using High Speed Camera, J. of Korean Institute of Information Technology, Vol.11, No.4, pp.1-6.

피인용 문헌

  1. Improvement of learning concrete crack detection model by weighted loss function vol.25, pp.10, 2018, https://doi.org/10.9708/jksci.2020.25.10.015
  2. 이미지 처리기법 및 레이저 센서를 이용한 휴대용 콘크리트 균열 측정 장치 개발에 관한 연구 vol.19, pp.4, 2020, https://doi.org/10.12814/jkgss.2020.19.4.041
  3. 영상을 활용한 시설물 안전점검 작업 효율성 향상 방안 연구 vol.39, pp.3, 2018, https://doi.org/10.7848/ksgpc.2021.39.3.179