Cloud Detection and Restoration of Landsat-8 using STARFM

재난 모니터링을 위한 Landsat 8호 영상의 구름 탐지 및 복원 연구

Lee, Mi Hee;Cheon, Eun Ji;Eo, Yang Dam

  • Received : 2019.09.20
  • Accepted : 2019.10.23
  • Published : 2019.10.31


Landsat satellite images have been increasingly used for disaster damage analysis and disaster monitoring because they can be used for periodic and broad observation of disaster damage area. However, periodic disaster monitoring has limitation because of areas having missing data due to clouds as a characteristic of optical satellite images. Therefore, a study needs to be conducted for restoration of missing areas. This study detected and removed clouds and cloud shadows by using the quality assessment (QA) band provided when acquiring Landsat-8 images, and performed image restoration of removed areas through a spatial and temporal adaptive reflectance fusion (STARFM) algorithm. The restored image by the proposed method is compared with the restored image by conventional image restoration method throught MLC method. As a results, the restoration method by STARFM showed an overall accuracy of 89.40%, and it is confirmed that the restoration method is more efficient than the conventional image restoration method. Therefore, the results of this study are expected to increase the utilization of disaster analysis using Landsat satellite images.


Landsat-8;cloud detection;image restoration;STARFM;disaster monitoring


Grant : 위성자료 활용 현업지원 기술개발

Supported by : 국립재난안전연구원