Journal of Convergence for Information Technology (융합정보논문지)

Convergence Society for SMB (중소기업융합학회)
권호 표지
DOI QR코드

DOI QR Code

Thermal Image Processing and Synthesis Technique Using Faster-RCNN

Faster-RCNN을 이용한 열화상 이미지 처리 및 합성 기법

  • Shin, Ki-Chul (Dept. of Electronic Computer Engineering, Inha University) ;
  • Lee, Jun-Su (Dept. of Mechanical Engineering, Inha University) ;
  • Kim, Ju-Sik (Dept. of Mechanical Engineering, Inha University) ;
  • Kim, Ju-Hyung (Dept. of Digital Solution Section Hydro&Nuclear Power Company) ;
  • Kwon, Jang-woo (Dept. of Electronic Computer Engineering, Inha University)
  • 신기철 (인하대학교 전기컴퓨터공학과) ;
  • 이준수 (인하대학교 기계공학과) ;
  • 김주식 (인하대학교 기계공학과) ;
  • 김주형 (한국수력원자력(주) 디지털솔루션부) ;
  • 권장우 (인하대학교 전기컴퓨터공학과)
  • Received : 2021.10.12
  • Accepted : 2021.12.20
  • Published : 2021.12.28
Download PDF
⟨ Previous Next ⟩

Abstract

In this paper, we propose a method for extracting thermal data from thermal image and improving detection of heating equipment using the data. The main goal is to read the data in bytes from the thermal image file to extract the thermal data and the real image, and to apply the composite image obtained by synthesizing the image and data to the deep learning model to improve the detection accuracy of the heating facility. Data of KHNP was used for evaluation data, and Faster-RCNN is used as a learning model to compare and evaluate deep learning detection performance according to each data group. The proposed method improved on average by 0.17 compared to the existing method in average precision evaluation.As a result, this study attempted to combine national data-based thermal image data and deep learning detection to improve effective data utilization.

본 논문에서는 열화상 이미지에서의 열 데이터 추출 및 해당 데이터를 사용한 발열 설비 탐지 향상 기법을 제안한다. 주요 목표는 열화상 이미지에서 바이트 단위로 데이터를 해석하여 열 데이터와 실화상 이미지를 추출하고 해당 이미지와 데이터를 합성한 합성 이미지를 딥러닝 모델에 적용하여 발열 설비의 탐지 정확도를 향상 시키는 것이다. 데이터는 한국수력원자력발전소 설비 데이터를 사용하였으며, 학습 모델로는 Faster-RCNN을 사용하여 각 데이터 그룹에 따른 딥러닝 탐지 성능을 비교 평가한다. 제안한 방식은 Average Precision 평가에서 기존 방식에 비해 평균 0.17 향상 되었다.본 연구는 이로서 국가 데이터 기반 열화상 데이터와 딥러닝 탐지의 접목을 시도하여 유효한 데이터 활용도 향상을 이루었다.

Keywords

Acknowledgement

본 논문은 KOREA HYDIRO&NUCLEAR POWER CO. LTD(No. 2018-Tech-07)의 지원을 받아 수행되었음

References

  1. J. Ma, Y. Ma & C. Li. (2019). Infrared and visible image fusion methods and applications: A survey. Information Fusion, (45), 153-178.
  2. B. B. Kang & J. Y. Kim. (2016). Anomaly diagnosis method using thermal image histogram signal characteristics. Proceedings of the Korean Urban Railway Association, 4(3), 521-528.
  3. K. H. Kim, S. S. Park, H. T. Lim, S. J. Lee & Y. S. Kim. (2019). Detecting leaks in old sewage pipes using thermal imaging information of an unmanned aerial system. Abstract of the Korean Society of Agricultural Engineering Conference, 159-159.
  4. Z. Guo, W. Liao, Y. Xiao, P. Veelaert & W. Philips. (2019). Deep learning fusion of RGB and depth images for pedestrian detection. The 30th British Machine Vision Conference (pp. 1-13). Cardiff : The BMVA.
  5. S. K. Lee, C. Y. Yoon, K. D. Moon, & C. G. Kim. (2020). A study on image synthesis-based training data augmentation for single-step detection method. Korean Information Processing Society, 27(1), 446-450.
  6. S. Ren, K. He, R. Girshick & J. Sun. (2015). Faster r-cnn: Towards real-time object detection with region proposal networks. Advances in neural information processing systems, 28, 91-99.
  7. W. Minkina & S. Dudzik. (2009). Infrared Thermography : Errors and Uncertainties. London : John Wiley & Sons
  8. P. Harvey. (n. d). ExifTool by Phil Harvey (Online). https://exiftool.org
  9. K. He, G. Gkioxari, P. Dollar & R. Girshick. (2017). Mask r-cnn. Proceedings of the IEEE international conference on computer vision (pp. 2961-2969). Cambridge : IEEE
  10. X. Wang, M. Yang, S. Zhu & Y. Lin. (2013). Regionlets for generic object detection. Proceedings of the IEEE international conference on computer vision (pp. 17-24). Cambridge : IEEE.
  11. H. J Kim, E. Dunn & J. M. Frahm. (2017). Learned contextual feature reweighting for image geo-localization. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 3251-3260). Cambridge : IEEE.