Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications
Cloud and cloud shadow (CCS) detection algorithms play a crucial role in the preprocessing of remote sensing data and directly affect the accuracy of subsequent analyses, making them an essential step in most analytical processes. Recent techniques for detecting CCS often employ deep learning method...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10833647/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1841536171812847616 |
|---|---|
| author | Keunhoo Cho Seongwook Park Boram Seong Seongwhan Lee Jae-Pil Park |
| author_facet | Keunhoo Cho Seongwook Park Boram Seong Seongwhan Lee Jae-Pil Park |
| author_sort | Keunhoo Cho |
| collection | DOAJ |
| description | Cloud and cloud shadow (CCS) detection algorithms play a crucial role in the preprocessing of remote sensing data and directly affect the accuracy of subsequent analyses, making them an essential step in most analytical processes. Recent techniques for detecting CCS often employ deep learning methods, which are effective but typically require extensive training data specific to each type of satellite imagery. This study presents a new methodology that applies a model trained on the preconstructed KOMPSAT-3/3A CCS dataset to Landsat-8 and Sentinel-2 satellite imagery. The experimental results demonstrated that the CCS detection model based on KOMPSAT-3/3A achieved a mean F1 score of 0.846 on the test dataset. When applied to Landsat-8 SPARCS and Sentinel-2 CloudSEN12 test datasets, it also showed high performance, with mean F1 scores of 0.741 and 0.8, respectively, effectively indicating that multi-modal CCS detection can be successfully implemented. Applying this model to different sensor imagery confirmed its effectiveness in gap filling, which can be utilized to enhance time-series analyses where continuous monitoring is required. In conclusion, this approach not only proves beneficial for time-series analysis but also significantly reduces the time and effort required to build datasets in deep learning-based CCS detection. |
| format | Article |
| id | doaj-art-4fcb2aff801f4ce58f03dd1f17b66e3e |
| institution | Kabale University |
| issn | 2169-3536 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj-art-4fcb2aff801f4ce58f03dd1f17b66e3e2025-01-15T00:02:34ZengIEEEIEEE Access2169-35362025-01-01137396740610.1109/ACCESS.2025.352695610833647Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling ApplicationsKeunhoo Cho0https://orcid.org/0000-0002-7847-982XSeongwook Park1https://orcid.org/0000-0002-2182-5322Boram Seong2Seongwhan Lee3Jae-Pil Park4Nara Space Technology Inc., Yeongdeungpo-gu, Seoul, South KoreaNara Space Technology Inc., Yeongdeungpo-gu, Seoul, South KoreaNara Space Technology Inc., Yeongdeungpo-gu, Seoul, South KoreaNara Space Technology Inc., Yeongdeungpo-gu, Seoul, South KoreaNara Space Technology Inc., Yeongdeungpo-gu, Seoul, South KoreaCloud and cloud shadow (CCS) detection algorithms play a crucial role in the preprocessing of remote sensing data and directly affect the accuracy of subsequent analyses, making them an essential step in most analytical processes. Recent techniques for detecting CCS often employ deep learning methods, which are effective but typically require extensive training data specific to each type of satellite imagery. This study presents a new methodology that applies a model trained on the preconstructed KOMPSAT-3/3A CCS dataset to Landsat-8 and Sentinel-2 satellite imagery. The experimental results demonstrated that the CCS detection model based on KOMPSAT-3/3A achieved a mean F1 score of 0.846 on the test dataset. When applied to Landsat-8 SPARCS and Sentinel-2 CloudSEN12 test datasets, it also showed high performance, with mean F1 scores of 0.741 and 0.8, respectively, effectively indicating that multi-modal CCS detection can be successfully implemented. Applying this model to different sensor imagery confirmed its effectiveness in gap filling, which can be utilized to enhance time-series analyses where continuous monitoring is required. In conclusion, this approach not only proves beneficial for time-series analysis but also significantly reduces the time and effort required to build datasets in deep learning-based CCS detection.https://ieeexplore.ieee.org/document/10833647/Cloudscloud shadowsmulti-modal imageryKOMPSAT-3KOMPSAT-3ASentinel-2 |
| spellingShingle | Keunhoo Cho Seongwook Park Boram Seong Seongwhan Lee Jae-Pil Park Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications IEEE Access Clouds cloud shadows multi-modal imagery KOMPSAT-3 KOMPSAT-3A Sentinel-2 |
| title | Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications |
| title_full | Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications |
| title_fullStr | Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications |
| title_full_unstemmed | Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications |
| title_short | Cloud and Cloud Shadow Detection for Multi-Modal Imagery With Gap-Filling Applications |
| title_sort | cloud and cloud shadow detection for multi modal imagery with gap filling applications |
| topic | Clouds cloud shadows multi-modal imagery KOMPSAT-3 KOMPSAT-3A Sentinel-2 |
| url | https://ieeexplore.ieee.org/document/10833647/ |
| work_keys_str_mv | AT keunhoocho cloudandcloudshadowdetectionformultimodalimagerywithgapfillingapplications AT seongwookpark cloudandcloudshadowdetectionformultimodalimagerywithgapfillingapplications AT boramseong cloudandcloudshadowdetectionformultimodalimagerywithgapfillingapplications AT seongwhanlee cloudandcloudshadowdetectionformultimodalimagerywithgapfillingapplications AT jaepilpark cloudandcloudshadowdetectionformultimodalimagerywithgapfillingapplications |