DOI QR코드

DOI QR Code

CycleGAN 기반 영상 모의를 적용한 건물지역 변화탐지 분석

The Analysis of Change Detection in Building Area Using CycleGAN-based Image Simulation

  • Jo, Su Min (Dept. of Technology Fusion Engineering, Konkuk University) ;
  • Won, Taeyeon (Dept. of Advanced Technology Fusion, Konkuk University, Realtimevisual Inc.) ;
  • Eo, Yang Dam (Dept. of Civil and Environmental Engineering, Konkuk University) ;
  • Lee, Seoungwoo (Dept. of Technology Fusion Engineering, Konkuk University)
  • 투고 : 2022.08.10
  • 심사 : 2022.08.24
  • 발행 : 2022.08.31

초록

원격탐사 영상의 변화탐지는 카메라의 광학적 요인, 계절적 요인, 토지피복 특성에 의해 오류가 발생한다. 본 연구에서는 CycleGAN (Cycle Generative Adversarial Network) 방법을 사용하여 촬영 각도에 따른 영상 내 건물 기울기를 모의 조정하였고, 이렇게 모의한 영상을 변화탐지에 활용하여 탐지 정확도 향상에 기여하도록 하였다. CycleGAN 기반으로 두 개 시기 영상 중 한 시기 영상을 기준으로 건물의 기울기를 다른 한 영상 내 건물에 유사하게 모의하였고 원 영상과 건물 기울기에 대한 오류를 비교 분석하였다. 실험자료로는 서로 다른 시기에 다른 각도로 촬영되었고, 건물이 밀집한 도시지역을 포함한 Kompsat-3A 고해상도 위성영상을 사용하였다. 실험 결과, 영상 내 건물 영역에 대하여 두 영상의 건물에 의한 오탐지 화소 수가 원 영상에서는 12,632개, CycleGAN 기반 모의 영상에서는 1,730개로 약 7배 감소하는 것으로 나타났다. 따라서, 제안 방법이 건물 기울기로 인한 탐지오류를 감소시킬 수 있음을 확인하였다.

The change detection in remote sensing results in errors due to the camera's optical factors, seasonal factors, and land cover characteristics. The inclination of the building in the image was simulated according to the camera angle using the Cycle Generative Adversarial Network method, and the simulated image was used to contribute to the improvement of change detection accuracy. Based on CycleGAN, the inclination of the building was similarly simulated to the building in the other image based on the image of one of the two periods, and the error of the original image and the inclination of the building was compared and analyzed. The experimental data were taken at different times at different angles, and Kompsat-3A high-resolution satellite images including urban areas with dense buildings were used. As a result of the experiment, the number of incorrect detection pixels per building in the two images for the building area in the image was shown to be reduced by approximately 7 times from 12,632 in the original image and 1,730 in the CycleGAN-based simulation image. Therefore, it was confirmed that the proposed method can reduce detection errors due to the inclination of the building.

키워드

과제정보

이 성과는 정부(과학기술정보통신부)의 재원으로 한국연구재단의 지원을 받아 수행된 연구임(No. 2019R1A2C1085618).

참고문헌

  1. Jang, Y.J., Oh, J.H., and Lee, C.N, (2020), Urban Building Change Detection Using nDSM and Road Extraction, Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, Vol. 38, No. 3, pp. 237-246. (in Korean with English abstract) https://doi.org/10.7848/KSGPC.2020.38.3.237
  2. Jensen, J. R., (2005), Introductory Digital Image Processing: A Remote Sensing Perspective 3rd Edition, Pearson, Prentice Hall, Upper Saddle River, NJ.
  3. Jung, S., Park, J., Lee, W.H., and Han, Y. (2020), Object-based building change detection using azimuth and elevation angles of sun and platform in the multi-sensor images, Korean Journal of Remote Sensing, Vol. 36, No. 5-2, pp. 989-1006. (in Korean with English abstract) https://doi.org/10.7780/kjrs.2020.36.5.2.12
  4. Mo, J., Seong, S., and Choi, J. (2021), Change Detection of Building Objects in Urban Area by Using Transfer Learning, Korean Journal of Remote Sensing, Vol. 37, No. 6-1, pp. 1685-1695. (in Korean with English abstract) https://doi.org/10.7780/KJRS.2021.37.6.1.16
  5. Park, S. and Song, A. (2022), Building change detection in high spatial resolution images using deep learning and graph model, Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, Vol. 40, No.3, pp. 227-237. (in Korean with English abstract) https://doi.org/10.7848/ksgpc.2022.40.3.227
  6. Shin, D., Kim, T., Han, Y., Kim, S., and Park, J. (2019), Change Detection of Building Demolition Area Using UAV, Korean Journal of Remote Sensing, Vol. 35, No. 5-2, pp. 819-829. (in Korean with English abstract) https://doi.org/10.7780/KJRS.2019.35.5.2.6
  7. Won, T., Jo, S.M., and Eo, Y.D., (2022), The Effect of Training Patch Size and ConvNeXt application on the Accuracy of CycleGAN-based Satellite Image Simulation, Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography, Vol. 40, No. 3, pp. 177-185. (in Korean with English abstract) https://doi.org/10.7848/ksgpc.2022.40.3.177
  8. Zhu, J.Y., Park, T., Isola, P., and Efros, A.A. (2017), Unpaired image-to-image translation using cycle-consistent adversarial networks, International Conference on Computer Vision, pp. 2223-2232. https://doi.org/10.48550/arXiv.1703.10593