All Issue

2026 Vol.36, Issue 4 Preview Page

Original Article

Analysis of RGB-CIELAB Color Distributions of Granite Cores and Comparison of Deep Learning-Based Weathering-Grade Prediction Performance

화강암 시추코어의 RGB-CIELAB 색상값 분포 분석과 딥러닝 기반 풍화등급 예측 성능 비교

Jun-Yeop Lee1
,
Hoyoung Jeong2*
이 준엽1
,
정 호영2*
1B.S., Department of Energy Resources Engineering, Pukyong National University
2Associate Professor, Department of Energy Resources Engineering, Pukyong National University
1국립부경대학교 에너지자원공학과 공학사
2국립부경대학교 에너지자원공학과 부교수
*Corresponding Author
31 August 2026. pp. 398-423
PDF XML Citation
Abstract

This study quantified the RGB, HSV, and CIELAB color distributions of granite drill-core images and compared the weathering-grade classification performance of support vector machines (SVMs) using quantile-based color features with that of ResNet-18-based deep-learning models. A large image dataset acquired under dry and wet conditions and three illumination levels was used for the color analysis, and the color characteristics associated with imaging conditions and weathering grades were examined using descriptive statistics. The classification experiments used 8,057 images obtained from 2,297 physical cores. The data were group-split into training, validation, and test sets at the physical-core level to ensure that images derived from the same core were not assigned to different subsets. The classification framework consisted of color-feature-based SVMs, a baseline ResNet-18 model using RGB tiles as input (R1), and a fusion model combining the core-level prediction probabilities of R1 with those of the color-feature-based SVM (R2). The fusion coefficient was determined using the validation set only. The test set was not used for model, checkpoint, or fusion-coefficient selection and was reserved for final prediction and a prespecified sensitivity analysis.The results showed that the deep-learning models outperformed the SVM models, with R2, which combined the RGB image model with an SVM incorporating CIELAB features, achieving the highest accuracy of 74.64%. These findings suggest that CIELAB-based color-distribution features may partially complement the predictive information learned by the RGB image model. However, further validation using independent field datasets is required to assess the generalizability and practical applicability of the proposed method.

Keywords
Rock weathering grade
Granite core
CIELAB
Deep learning

본 연구에서는 화강암 시추코어 이미지의 RGB, HSV 및 CIELAB 색상분포를 정량화하고, 분위수 기반 색상특징을 이용한 SVM과 ResNet-18 기반 딥러닝 모델의 풍화등급 분류성능을 비교하였다. 색상분석에는 건조·습윤 상태와 세 가지 조도조건에서 촬영된 대규모 이미지 자료를 사용하였으며, 촬영조건과 풍화등급에 따른 색상특성을 기술통계적으로 분석하였다. 분류실험에는 2,297개 물리 코어에서 획득한 8,057장의 이미지를 사용하였고, 동일 물리 코어에서 생성된 이미지가 서로 다른 자료군에 포함되지 않도록 물리 코어 단위로 학습, 검증 및 시험자료를 그룹 분할하였다. 분류모델은 색상특징 기반 SVM, RGB 타일을 입력으로 사용하는 ResNet-18 기준모델(R1), 그리고 R1의 코어 단위 예측확률과 색상특징 SVM의 예측확률을 결합한 융합모델(R2)로 구성하였다. 결합계수는 검증자료만을 이용하여 결정하였으며, 시험자료는 최종 성능평가에 1회만 사용하였다. 시험 결과, SVM 모델보다 딥러닝 모델의 분류성능이 높게 나타났으며, CIELAB 특징을 포함한 SVM과 RGB 이미지 기반 모델을 결합한 R2가 74.64%의 정확도로 가장 우수한 성능을 보였다. 이러한 결과는 CIELAB 기반 색상분포 특징이 RGB 이미지 기반 모델의 예측정보를 부분적으로 보완할 가능성을 보여준다. 다만, 제안 방법의 일반화 가능성과 현장 적용성을 확인하기 위해서는 독립된 현장자료를 이용한 추가 검증이 필요하다.

키워드
암석 풍화등급
화강암 코어
CIELAB
딥러닝
References
1

AI-Hub, 2024, 기반암 시추 시료를 이용한 암반 등급 분류 데이터, National Information Society Agency, https://www.aihub.or.kr/aihubdata/data/view.do?aihubDataSe=data&currMenu=120&dataSetSn=71753 (accessed July 28, 2026).

2

Alavi Nezhad Khalil Abad, S.V., Tugrul, A., Gokceoglu, C., and Jahed Armaghani, D., 2016, Characteristics of weathering zones of granitic rocks in Malaysia for geotechnical engineering design, Engineering Geology, 200, 94-103.

10.1016/j.enggeo.2015.12.006
3

Baek, S.H., Park, K.H., Jeon, J.S., and Kwak, T.Y., 2023, A novel method for calibration of digital soil images captured under irregular lighting conditions, Sensors, 23(1), 296.

10.3390/s2301029636616893PMC9824125
4

Cohen, J., 1960, A coefficient of agreement for nominal scales, Educational and Psychological Measurement, 20(1), 37-46.

10.1177/001316446002000104
5

Cohen, J., 1968, Weighted kappa: Nominal scale agreement with provision for scaled disagreement or partial credit, Psychological Bulletin, 70(4), 213-220.

10.1037/h0026256
6

Cortes, C. and Vapnik, V., 1995, Support-vector networks, Machine Learning, 20, 273-297.

10.1023/A:1022627411411
7

Dearman, W.R., Baynes, F.J., and Irfan, T.Y., 1978, Engineering grading of weathered granite, Engineering Geology, 12, 345-374.

10.1016/0013-7952(78)90018-2
8

Efron, B. and Tibshirani, R.J., 1993, An Introduction to the Bootstrap, Chapman and Hall, New York.

10.1007/978-1-4899-4541-9
9

He, K., Zhang, X., Ren, S., and Sun, J., 2016, Deep residual learning for image recognition, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770-778.

10.1109/CVPR.2016.90
10

Lazzerini, B. and Pistolesi, F., 2020, Evaluation of the role of image resolution in the perception of color differences, Proceedings of IEEE International Joint Conference on Neural Networks (IJCNN 2020), part of the IEEE World Congress on Computational Intelligence (WCCI 2020), Glasgow, UK.

10.1109/IJCNN48605.2020.9207282
11

Mohamad Nasir, N. and Misro, M.Y., 2024, Analysis of rock mass weathering grade using image processing technique, Malaysian Journal of Fundamental and Applied Sciences, 20(3), 544-560.

10.11113/mjfas.v20n3.3344
12

Nagano, T. and Nakashima, S., 1989, Study of colors and degrees of weathering of granitic rocks by visible diffuse reflectance spectroscopy, Geochemical Journal, 23(2), 75-83.

10.2343/geochemj.23.75
13

Nicolás-Sáenz, L., Ledezma, A., Pascau, J., and Muñoz-Barrutia, A., 2023, ABANICCO: A New Color Space for Multi-Label Pixel Classification and Color Analysis, Sensors, 23(6), 3338.

10.3390/s2306333836992044PMC10052715
14

Pham, C. and Shin, H.S., 2020, A feasibility study on application of a deep convolutional neural network for automatic rock type classification, Tunnel and Underground Space, 30(5), 462-472.

10.7474/TUS.2020.30.5.462
15

Platt, J., 1999, Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods, Advances in Large Margin Classifiers, MIT Press, pp. 61-74.

10.7551/mitpress/1113.003.0008
16

Popov, V., Ostarek, M., and Tenison, C., 2018, Practices and pitfalls in inferring neural representations, NeuroImage, 174, 340-351.

10.1016/j.neuroimage.2018.03.041
17

Potočnik, M., Klemenc, B., Solina, F., and Herlec, U., 2015, Computer aided method for colour calibration and analysis of digital rock photographs, Geologija, 58(2), 247-260.

10.5474/geologija.2015.020
18

Tse, K.K.C., Wong, L.N.Y., Cheung, S.H., and Yu, L., 2025, Advanced classification of drill core rock type and weathering grade using detection transformer-based artificial intelligence techniques, Journal of Rock Mechanics and Geotechnical Engineering, 17(7), 4036-4045.

10.1016/j.jrmge.2024.09.016
19

Yusoff, I.N., Mohamad Ismail, M.A., Tobe, H., Date, K., and Yokota, Y., 2023, Quantitative granitic weathering assessment for rock mass classification optimization of tunnel face using image analysis technique, Ain Shams Engineering Journal, 14(1), 101814.

10.1016/j.asej.2022.101814
Information
  • Publisher :Korean Society for Rock Mechanics and Rock Engineering
  • Publisher(Ko) :한국암반공학회
  • Journal Title :Tunnel and Underground Space
  • Journal Title(Ko) :터널과 지하공간
  • Volume : 36
  • No :4
  • Pages :398-423
  • Received Date : 2026-08-04
  • Revised Date : 2026-08-27
  • Accepted Date : 2026-08-28
  • DOI :https://doi.org/10.7474/TUS.2026.36.4.398
Journal Informaiton Tunnel and Underground Space Tunnel and Underground Space
Online Submission
Archives
: Vol.27, 2017 ~
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close