All Issue

2024 Vol.26, Issue 6

Technical Paper

Development of an automatic region of interest setting algorithm based on tunnel CCTV

터널 CCTV 기반 관심영역 자동설정 알고리즘 개발

Kyu Beom Lee1
,
Hyu-Soung Shin2*
이 규범1
,
신 휴성2*
1Student Researcher, Department of Future & Smart Construction Research, Korea Institute of Civil Engineering and Building Technology / Ph.D. Student, Smart City and Construction Convergence, University of Science & Technology
2Senior Research Fellow, Department of Future & Smart Construction Research, Korea Institute of Civil Engineering and Building Technology / CEO, SIMAN Ltd.
1학생회원, 한국건설기술연구원 미래스마트건설연구본부 학생연구원 / 과학기술연합대학원대학교(UST) 스마트도시건설융합 석박통합과정
2정회원, 한국건설기술연구원 미래스마트건설연구본부 선임연구위원 / (주)시만 대표이사
*Corresponding Author
30 November 2024. pp. 563-582
PDF XML Citation
Abstract

The existing tunnel CCTV-based accident detection system has a problem where false alarms occur as the pre-set Region of interest (ROI) fails to detect incidents properly when the tunnel CCTV rotates. To address this issue, this paper develops an Automatic ROI setting algorithm that automatically reconfigures the ROI in response to the rotation of the tunnel CCTV. This algorithm first sets the ROI in the original tunnel CCTV image and then applies an Inverse perspective transform (IPT) to generate a transformed image with the perspective effect removed. In the image, a lane detection algorithm extracts the slope, intercept, and direction (left/center/right) of each lane, and based on this information, the ROI is automatically reconfigured in the original image to match the rotated CCTV. To validate this, we first conducted experiments on a deep learning-based lane detection model and created datasets using the lower half and lower quarter of still images for comparison. The results showed that the model using the lower half performed robustly even when the road surface was wet or the lane markings were blurred. Based on these results, the developed algorithm was applied, and a comparative experiment was conducted by selecting three CCTV channels, with and without the application of the algorithm. The results confirmed that the algorithm consistently maintained lane direction compared to the existing fixed ROI, and corrected lanes tilted up to 37 degrees to ensure a consistent ROI. Consequently, the proposed algorithm has been proven to significantly enhance the accuracy and efficiency of the system.

Keywords
Automatic region of interest setting algorithm
Lane detection algorithm
Inverse perspective transform
Transformed image
Region of interest

기존 터널 영상 유고 시스템은 터널 내 고압수 청소 등으로 인해 빈번히 CCTV가 틀어지면서 미리 설정된 관심영역과 실제 차량 주행방향과의 차이가 발생하면서 유고 상황을 제대로 감지하지 못하고 오탐지가 발생하는 문제가 빈번히 발생한다. 이러한 문제를 해결하기 위해, 본 논문에서는 터널 CCTV의 틀어짐에 대응하여 관심영역을 자동으로 재설정하는 관심영역 자동설정 알고리즘을 개발하였다. 이 알고리즘은 원본 터널 CCTV 영상에서 관심영역을 설정한 후, 역 원근변환(inverse perspective transform, IPT)을 적용하여 원근현상이 제거된 변환영상을 기본 영상으로 사용한다. 변환영상에서 차선 인식 알고리즘을 통해 각 차선의 기울기, 절편, 방향(좌/중앙/우)을 추출하고, 이를 바탕으로 원본영상에서 틀어진 CCTV에 맞추어 관심영역을 자동으로 재설정한다. 이에 대한 성능을 검증하기 위해 먼저 딥러닝 기반 차선 인식 모델에 대한 실험을 진행하였으며, 정지 영상의 하단 1/2 및 1/4 영역을 사용한 데이터셋을 제작하여 비교 실험을 수행하였다. 그 결과, 하단 1/2 영역을 활용한 모델이 노면이 젖은 경우나 차선이 흐릿한 경우에도 높은 차선인식 성능을 보였다. 이러한 결과를 바탕으로 개발된 알고리즘에 적용하였으며, CCTV 3개 채널을 선정한 다음, 본 알고리즘을 적용한 경우와 그렇지 않은 경우에 대한 비교 실험을 진행하였다. 그 결과, 본 알고리즘은 기존의 고정된 관심영역에 비해 차선 방향을 일관되게 유지할 수 있었으며, 최대 37도까지 틀어진 차선을 자동으로 보정하여 일관된 관심영역을 재설정할 수 있음을 확인하였다. 결과적으로 제안된 알고리즘은 터널 영상유고시스템의 정확도와 효율성을 크게 향상시킬 수 있음을 보였다.

키워드
관심영역 자동설정 알고리즘
차선인식 알고리즘
역 원근변환
변환영상
관심영역
References
1

Cao, J., Song, C., Song, S., Xiao, F., Peng, S. (2019), "Lane detection algorithm for intelligent vehicles in complex road conditions and dynamic environments", Sensors, Vol. 19, No. 14, 3166.

10.3390/s1914316631323875PMC6679325
2

Cordts, M., Omran, M., Ramos, S., Rehfeld, T., Enzweiler, M., Benenson, R., Franke, U., Roth, S., Schiele, B. (2016), "The cityscapes dataset for semantic urban scene understanding", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, pp. 3213-3223.

10.1109/CVPR.2016.350
3

Fischler, M.A., 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
4

Galeano, D.A.B., Devy, M., Boizard, J.L., Filali, W. (2011), "Real-time architecture on FPGA for obstacle detection using inverse perspective mapping", Proceedings of the 18th IEEE International Conference on Electronics, Circuits, and Systems, Beirut, pp. 788-791.

10.1109/ICECS.2011.6122392
5

He, K., Gkioxari, G., Dollár, P., Girshick, R. (2017), "Mask R-CNN", Proceedings of the IEEE International Conference on Computer Vision, Venice, pp. 2961-2969.

10.1109/ICCV.2017.322
6

LeCun, Y., Bengio, Y., Hinton, G. (2015), "Deep learning", Nature, Vol. 521, No. 7553, pp. 436-444.

10.1038/nature1453926017442
7

Lee, K.B., Shin, H.S. (2022), "A study for improvement of far-distance performance of a tunnel accident detection system by using an inverse perspective transformation", Journal of Korean Tunnelling and Underground Space Association, Vol. 24, No. 3, pp. 247-262.

10.9711/KTAJ.2022.24.3.247
8

Lee, K.B., Shin, H.S. (2023), "Study on influence of the region of interest (ROI) in performance of accident event detection in tunnel CCTV", Journal of the Korea Academia-Industrial Cooperation Society, Vol. 24, No. 2, pp. 619-624.

10.5762/KAIS.2023.24.2.619
9

Mallot, H.A., Bülthoff, H.H., Little, J.J., Bohrer, S. (1991), "Inverse perspective mapping simplifies optical flow computation and obstacle detection", Biological Cybernetics, Vol. 64, No. 3, pp. 177-185.

10.1007/BF002019782004128
10

MOLIT (2021), Guideline of installation and management of disaster prevention facilities on road tunnels (established rule of MOLIT No. 336), Ministry of Land, Infrastructure and Transport.

11

Ren, S., He, K., Girshick, R., Sun, J. (2015), "Faster R-CNN: towards real-time object detection with region proposal networks", Proceedings of the Advances in Neural Information Processing Systems 28, Montreal, pp. 91-99.

12

Sultana, S., Ahmed, B., Paul, M., Islam, M.R., Ahmad, S. (2023), "Vision-based robust lane detection and tracking in challenging conditions", IEEE Access, Vol. 11, pp. 67938-67955.

10.1109/ACCESS.2023.3292128
13

Zhu, M. (2004), "Recall, precision and average precision", Working Paper, Department of Statistics and Actuarial Science, University of Waterloo, Vol. 2, No. 30, pp. 1-11.

Information
  • Publisher :Korean Tunneling and Underground Space Association
  • Publisher(Ko) :한국터널지하공간학회
  • Journal Title :Journal of Korean Tunnelling and Underground Space Association
  • Journal Title(Ko) :한국터널지하공간학회 논문집
  • Volume : 26
  • No :6
  • Pages :563-582
  • Received Date : 2024-09-23
  • Revised Date : 2024-10-22
  • Accepted Date : 2024-10-28
  • DOI :https://doi.org/10.9711/KTAJ.2024.26.6.563
Journal Informaiton Journal of Korean Tunnelling and Underground Space Association Journal of Korean Tunnelling and Underground Space Association
Online Submission & Review System
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • crosscheck
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close