All Issue

2025 Vol.59, Issue 4 Preview Page

Research Article

Deep Learning–Based Detection of Marine Oil Spill from SAR Images

SAR 영상과 딥러닝을 이용한 해상 유출유 탐지

Jonggu Kanga
,
Seungrok Leeb
,
Yangwon Leec
,
Daesun Kimd*
강 종구a
,
이 승록b
,
이 양원c
,
김 대선d*
aPh.D. Student, Department of Spatial Information Engineering, Pukyong National University
bSenior Researcher, Forecasting Business Division, Geosystem Research Co., Ltd.
cProfessor, Department of Geomatics, Pukyong National University
dSenior Researcher, Korea Institute of Ocean Science and Technology
a국립부경대학교 지구환경시스템과학부 공간정보공학전공 박사과정
b(주)지오시스템리서치 예보사업부 선임
c국립부경대학교 지구환경시스템과학부 위성정보융합공학전공 교수
d교신저자, 한국해양과학기술원 선임연구원
*dskim877@kiost.ac.kr
31 December 2025. pp. 377-387
PDF XML Citation
Abstract

As human activities in marine environments increase, the risk of oil spills from maritime accidents continues to rise. Accurate situational information is essential for rapid response and effective mitigation to minimize damage. Satellite-based monitoring provides superior spatial and temporal coverage compared to localized observation systems, while synthetic aperture radar (SAR) imagery enables all-weather, day-and-night observations. Recent advances in deep learning–based image recognition have shown that Transformer models effectively detect diverse oil spill patterns under complex marine conditions. In this study, a Transformer-based segmentation model was constructed using a publicly available Sentinel-1 SAR dataset. Especially, Hyperparameter optimization was performed with the Optuna framework, followed by a blind evaluation using random sampling. The proposed model achieved an accuracy of 0.992, precision of 0.909, recall of 0.865, F1-score of 0.887, and an intersection over union (IoU) of 0.797. Despite speckle noise and sea surface reflectivity variations, the model accurately and precisely delineated oil spill regions. Stable performance was also observed in scenes with complex oil slick boundaries, producing segmentation results closely matching the reference labels.

Keywords
Oil spill detection
Deep learning
SAR Imagery

해양공간의 인간활동의 증가로 해상사고에 의한 유출유 발생위험은 점점 커지고 있으며, 신속한 방제로 유출유 피해를 최소화하기 위해서는 정확한 현황정보 수집이 매우 중요하다. 위성영상을 통한 유출유 모니터링은 국지적 활용장비에 비해 시공간적으로 효율적이며, 특히 SAR 영상은 주야간 전천후 관측이 가능하다는 장점을 가진다. 최근 딥러닝 영상인식 분야에서는 복잡한 해양환경에서 다양한 유출유 패턴을 효과적으로 탐지하는 Transformer 모델이 우수한 성능으로 평가되고 있다. 이에 본 연구에서는 Sentinel-1을 통해 구축된 공개 SAR 데이터셋을 활용하여 Transformer 기반 세그멘테이션 모델을 구축하여 해상 유출유를 탐지하였다. 특히 Optuna 프레임워크를 통해 하이퍼파라미터 최적화를 수행하고 공식 데이터 분할을 통한 암맹평가를 수행한 결과, Accuracy는 0.992, Precision은 0.909, Recall은 0.865, F1-score는 0.887, IoU는 0.797로 나타났다. 특히 해양 표면의 스펙클 패턴이나 반사 강도 변화에도 불구하고 유출유 영역을 비교적 정확하고 세밀하게 분리하였으며, 복잡한 유막 경계를 포함한 장면에서도 실제 라벨과 유사한 형태를 안정적으로 복원하는 성능을 나타냈다.

키워드
해상 유출유 탐지
딥러닝
SAR 영상
References
1

강종구･윤유정･김근아･박강현･최소연･양찬수･이종혁･이양원, 2022, “DeepLabV3+ 모델을 이용한 PlanetScope 영상의해상 유출유 탐지,” Korean Journal of Remote Sensing 38(6): 1623-1631.

2

강종구･이양원･김대선, 2023, “Sentinel-2 위성과 국토위성에서의 딥러닝 기반 초해상화 기법 비교 연구,” 국토지리학회지 57(4): 541-555.

10.22905/kaopqj.2023.57.4.13
3

공공데이터포털, 2025, 해양경찰청 해양오염사고 상세데이터 현황.

4

김태호･신혜경･장소영･유정미･김평중･양찬수, 2021, “고해상도 광학위성을 이용한 해상 유출유 면적 산출: 심포니호 기름유출 사고 사례,” Korean Journal of Remote Sensing, 37(6-1): 1773-1784.

5

Alpers, W., B. Holt, and K. Zeng, 2017, Oil spill detection by imaging radars: Challenges and pitfalls, Remote Sensing of Environment 201: 133-147. https://doi.org/10.1016/j.rse.2017.09.002

10.1016/j.rse.2017.09.002
6

Arslan, N., M. Majidi Nezhad, A. Heydari, D. Astiaso Garcia, and G. Sylaios, 2023, A principal component analysis methodology of oil spill detection and monitoring using satellite remote sensing sensors, Remote Sensing 15(5): 1460.

10.3390/rs15051460
7

Aznar, F., M. Sempere, M. Pujol, R. Rizo, and M. J. Pujol, 2014, Modelling oil-spill detection with swarm drones, Abstract and Applied Analysis 2014: 949407. https://doi.org/10.1155/2014/949407

10.1155/2014/949407
8

Dong, S., J. Feng, Z. Gu, K. Yin, and Y. Long, 2025, A review of artificial intelligence and remote sensing for marine oil spill detection, classification, and thickness estimation, Remote Sensing 17(22): 3681.

10.3390/rs17223681
9

Hong, X., L. Chen, S. Sun, Z. Sun, Y. Chen, Q. Mei, and Z. Chen, 2022, Detection of oil spills in the northern South China Sea using Landsat-8 OLI, Remote Sensing 14(16): 3966.

10.3390/rs14163966
10

Huang, X., B. Zhang, W. Perrie, Y. Lu, and C. Wang, 2022, A novel deep learning method for marine oil spill detection from satellite synthetic aperture radar imagery, Marine Pollution Bulletin 179: 113666.

10.1016/j.marpolbul.2022.113666
11

Kang, J., C. Yang, J. Yi, and Y. Lee, 2024, Detection of marine oil spill from planetscope images using CNN and transformer models, Journal of Marine Science and Engineering 12(11): 2095.

10.3390/jmse12112095
12

Kim, D. and H. S. Jung, 2017, Oil spill detection from RADARSAT-2 SAR image using non-local means filter, Korean Journal of Remote Sensing 33(1): 61-67.

10.7780/kjrs.2017.33.1.6
13

Krestenitis, M., G. Orfanidis, K. Ioannidis, K. Avgerinakis, S. Vrochidis, and I. Kompatsiaris, 2019, Oil spill identification from satellite images using deep neural networks, Remote Sensing 11(15): 1762.

10.3390/rs11151762
14

Odonkor, P., Z. Ball, and S. Chowdhury, 2019, Distributed operation of collaborating unmanned aerial vehicles for time-sensitive oil spill mapping, Swarm and Evolutionary Computation 46: 52-68. https://doi.org/10.1016/j.swevo.2019.01.005

10.1016/j.swevo.2019.01.005
15

Schaeffer, B. A., P. Whitman, R. Conmy, W. Salls, M. Coffer, D. Graybill, and M. C. Lebrasse, 2022, Potential for commercial PlanetScope satellites in oil response monitoring, Marine pollution bulletin 183, 114077.

10.1016/j.marpolbul.2022.11407736084611PMC10034735
16

Solberg, A. H. S., 2012, Remote sensing of ocean oil-spill pollution, Proceedings of the IEEE 100(10): 2931-2945. https://doi.org/10.1109/JPROC.2012.2196250. 2196250

10.1109/JPROC.2012.2196250
17

Trujillo-Acatitla, R., J. Tuxpan-Vargas, C. Ovando-Vázquez, and E. Monterrubio-Martínez, 2024, Marine oil spill detection and segmentation in SAR data with two steps deep learning framework, Marine Pollution Bulletin 204: 116549.

10.1016/j.marpolbul.2024.116549
18

Xie, E., W. Wang, Z. Yu, A. Anandkumar, J. M. Alvarez, and P. Luo, 2021, SegFormer: Simple and efficient design for semantic segmentation with transformers. Advances in neural information processing systems 34: 12077-12090.

19

Zeng, K., and Y. Wang, 2020, A deep convolutional neural network for oil spill detection from spaceborne SAR images. Remote Sensing 12(6): 1015.

10.3390/rs12061015
20

Zhang, C., G. Gao, X. Zhang, H. C. Li, T. Liu, Z. Zhang, and G. Li, 2025, An overview of polarized synthetic aperture radar ship detection: From traditional methods to artificial intelligence, IEEE Geoscience and Remote Sensing Magazine, 13(4): 174-209.

10.1109/MGRS.2025.3572627
Information
  • Publisher :The Korean Association of Professional Geographers
  • Publisher(Ko) :국토지리학회
  • Journal Title :국토지리학회지
  • Journal Title(Ko) :THE GEOGRAPHICAL JOURNAL OF KOREA
  • Volume : 59
  • No :4
  • Pages :377-387
  • Received Date : 2025-12-17
  • Revised Date : 2025-12-22
  • Accepted Date : 2025-12-23
  • DOI :https://doi.org/10.22905/kaopqj.2025.59.4.7
Journal Informaiton 국토지리학회지 국토지리학회지
Online Submission
  • NRF
  • KISTI Current Status
  • KISTI Cited-by
  • KCI 문헌 유사도 검사
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close