All Issue

2026 Vol.59, Issue 4 Preview Page
Construction of a river image dataset and analysis of the SIFT algorithm

하천 이미지 데이터셋 구축 및 SIFT 알고리즘 분석

Seowon Joa
,
Jinyong Kimb
,
Donghwi Jungc*
조 서원a
,
김 진용b
,
정 동휘c*
aMaster’s Student, Department of Civil, Environmental and Architectural Engineering, College of Engineering, Korea University, Seoul, Korea
bResearcher, Future and Fusion Lab of Architectural, Civil and Environmental Engineering, Korea University, Seoul, Korea
cAssociate Professor, School of Civil, Environmental and Architectural Engineering, College of Engineering, Korea University, Seoul, Korea
a고려대학교 건축사회환경공학과 석사과정
b고려대학교 미래건설환경융합연구소 연구원
c고려대학교 건축사회환경공학부 부교수
*Corresponding Author
30 April 2026. pp. 349-359
PDF XML Citation
Abstract

Image-based river monitoring research has rapidly expanded through the utilization of diverse observational data sources, including satellites, unmanned aerial vehicles (UAVs), and CCTV systems. However, little effort has been devoted to systematically constructing and publicly sharing benchmark dataset for Korean rivers. Moreover, systematic verification of how environmental variables, such as illumination conditions, affect the keypoint detection and feature matching performance of feature-based algorithms in actual river environments has been limited. In this study, the KU River Image Dataset, comprising 1,732 images that reflect the spatiotemporal diversity across major watersheds in Korea, was constructed and publicly released. Analysis using SIFT, a keypoint detection method, was performed on some of the images in the dataset. The analysis revealed that, during daytime, the number and variability of keypoints were relatively high, whereas at nighttime, the number decreased but the keypoints exhibited more stable characteristics. In terms of algorithm performance, the highest feature matching rates were observed for nighttime-nighttime image pairs with stable illumination, and the optimal octave—an octave being the image resolution level used for keypoint detection—varied depending on illumination conditions. Daytime-nighttime image pairs exhibited relatively lower feature matching performance. This confirms that illumination is a key factor affecting image registration and feature matching in river images.

Keywords
River monitoring
Computer vision
Scale-Invariant Feature Transform (SIFT)
Illumination variation
Benchmark dataset

이미지 기반 하천 모니터링 연구는 위성, 무인항공기(UAV), CCTV 등 다양한 관측 자료를 활용하여 빠르게 확산되고 있다. 그러나 국내 하천을 대상으로 한 벤치마크 데이터셋이 부재할 뿐만 아니라, 실제 하천 환경에서 조도와 같은 환경 변수가 특징점 기반 알고리즘의 특징점 검출 및 특성 매칭 성능에 미치는 영향을 체계적으로 검증한 연구는 미비한 실정이다. 이에 본 연구에서는 국내 주요 유역의 시·공간적 다양성을 반영한 KU 하천 이미지 데이터셋(KU River Image Dataset) 1,732장을 구축·공개하였으며, 이 중 일부 이미지를 활용하여 특징점 검출 기법인 스케일 불변 특징 변환(SIFT) 기반의 분석을 수행하였다. 실험 결과, 주간에는 특징점의 개수가 증가하고 변동성이 상대적으로 크게 나타난 반면, 야간에는 개수가 감소하였으나 보다 안정적인 특성을 보였다. 더불어 알고리즘 성능 평가에서 조도가 안정적인 야간-야간 조합이 가장 높은 특징 매칭 성능을 기록하였으며, 조도 조건에 따라 최적 성능을 나타내는 이미지 해상도 단계(옥타브)가 상이함을 확인하였다. 또한, 주간-야간 교차 정합에서는 상대적으로 낮은 특징 매칭 성능을 보임으로써, 조도 조건이 하천 이미지 정합 및 특징 매칭에서 핵심적인 변수임을 확인하였다.

키워드
하천 모니터링
컴퓨터 비전
스케일 불변 특징 변환(SIFT)
조도 변화
벤치마크 데이터셋
References
1

Bentivoglio, R., Isufi, E., Jonkman, S.N., and Taormina, R. (2022). “Deep learning methods for flood mapping: A review of existing applications and future research directions.” Hydrology and Earth System Sciences, Vol. 26, pp. 4345-4378. doi: 10.5194/hess-26-4345-2022.

10.5194/hess-26-4345-2022
2

Bonafilia, D., Tellman, B., Anderson, T., and Issenberg, E. (2020). “Sen1Floods11: A georeferenced dataset to train and test deep learning flood algorithms for Sentinel-1.” Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, Seattle, WA, U.S., pp. 210-211.

10.1109/CVPRW50498.2020.00113
3

Brown, M., and Lowe, D.G. (2007). “Automatic panoramic image stitching using invariant features.” International Journal of Computer Vision, Vol. 74, No. 1, pp. 59-73.

10.1007/s11263-006-0002-3
4

Eltner, A., Sardemann, H., and Grundmann, J. (2020). “Flow velocity and discharge measurement in rivers using terrestrial and unmanned-aerial-vehicle imagery.” Hydrology and Earth System Sciences, Vol. 24, No. 3, pp. 1429-1445.

10.5194/hess-24-1429-2020
5

Fischler, M.A., and Bolles, R.C. (1981). “Random sample consensus: A paradigm for model fitting with applications to image analysis and automated cartography.” Communications of the ACM, Vol. 24, No. 6, pp. 381-395.

10.1145/358669.358692
6

Jun, S., Jang, H., Kim, S., Lee, J.S., and Jung, D. (2024). “A review of ground camera-based computer vision techniques for flood management.” Computers and Concrete, Vol. 33, No. 4, pp. 425-443.

10.12989/CAC.2024.33.4.425
7

Kim, J., Kim, S., and Jung, D. (2025). “Foundation model for waterbody segmentation based on ground camera images.” Journal of Korea Water Resources Association, Vol. 58, No. 6, pp. 449-457.

10.3741/JKWRA.2025.58.6.449
8

Liu, W., Wang, B., Guo, Y., and Sun, H. (2024). “Integrated monitoring method of flood free surface and surface velocity in a laboratory compound channel.” Journal of Hydrology, Vol. 635, 131149.

10.1016/j.jhydrol.2024.131149
9

Lowe, D.G. (2004). “Distinctive image features from scale-invariant keypoints.” International Journal of Computer Vision, Vol. 60, No. 2, pp. 91-110.

10.1023/B:VISI.0000029664.99615.94
10

Mikolajczyk, K., and Schmid, C. (2005). “A performance evaluation of local descriptors.” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 27, No. 10, pp. 1615-1630.

10.1109/TPAMI.2005.188
11

Peng, J., Huang, Y., Sun, W., Chen, N., Ning, Y., and Du, Q. (2022). “Domain adaptation in remote sensing image classification: A survey.” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Vol. 15, pp. 9842-9859.

10.1109/JSTARS.2022.3220875
12

Rahnemoonfar, M., Chowdhury, T., Sarkar, A., Varshney, D., Yari, M., and Murphy, R.R. (2021). “FloodNet: A high-resolution aerial imagery dataset for post-flood scene understanding.” IEEE Access, Vol. 9, pp. 89644-89654.

10.1109/ACCESS.2021.3090981
13

Stylianou, A., Abrams, A., and Pless, R. (2015). “Characterizing feature matching performance over long time periods.” Proceedings of the 2015 IEEE Winter Conference on Applications of Computer Vision, IEEE, Waikoloa, HI, U.S., pp. 892-898.

10.1109/WACV.2015.123
14

Szeliski, R. (2022). Computer vision: Algorithms and applications. Springer Nature, Cham, Switzerland.

10.1007/978-3-030-34372-9
15

The KU River Image Dataset in Zenodo (2026). published 10 February 2026, <https://doi.org/10.5281/zenodo.18593957>.

10.5281/zenodo.18593957>
16

Valgren, C., and Lilienthal, A.J. (2007). “Sift, surf and seasons: Long-term outdoor localization using local features.” Proceedings of the 3rd European Conference on Mobile Robots, Freiburg, Germany, pp. 253-258.

Information
  • Publisher :KOREA WATER RESOURECES ASSOCIATION
  • Publisher(Ko) :한국수자원학회
  • Journal Title :Journal of Korea Water Resources Association
  • Journal Title(Ko) :한국수자원학회 논문집
  • Volume : 59
  • No :4
  • Pages :349-359
  • Received Date : 2025-12-23
  • Revised Date : 2026-02-10
  • Accepted Date : 2026-02-20
  • DOI :https://doi.org/10.3741/JKWRA.2026.59.4.349
Journal Informaiton Journal of Korea Water Resources Association Journal of Korea Water Resources Association
Online Submission
Archives
  • scopus
  • KCI-수자원
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close