Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery
Cloud and cloud shadow detection is one of the most important tasks for optical remote sensing image preprocessing. It is not an easy task due to the variety and complexity of underlying surfaces, such as the low-albedo objects (water and mountain shadow) and the high-albedo objects (snow and ice)....
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/8962046/ |
| _version_ | 1818929845729492992 |
|---|---|
| author | Yang Chen Luliang Tang Zihan Kan Aamir Latif Xiucheng Yang Muhammad Bilal Qingquan Li |
| author_facet | Yang Chen Luliang Tang Zihan Kan Aamir Latif Xiucheng Yang Muhammad Bilal Qingquan Li |
| author_sort | Yang Chen |
| collection | DOAJ |
| description | Cloud and cloud shadow detection is one of the most important tasks for optical remote sensing image preprocessing. It is not an easy task due to the variety and complexity of underlying surfaces, such as the low-albedo objects (water and mountain shadow) and the high-albedo objects (snow and ice). In this study, an end-to-end multiscale 3D-CNN method is proposed for cloud and cloud shadow detection in high resolution multispectral imagery. Specifically, a multiscale learning module is designed to extract cloud and cloud shadow contextual information of different levels. In order to make full use of band information, four band-combination images are inputted into the multiscale 3D-CNN. A joint spectral-spatial information of 3D-convolution layer is developed to fully explore the joint spatial-spectral correlations feature in the input data. Overall, in the experiments undertaken in this paper, the proposed method achieved a mean overall accuracy of 97.27% for cloud detection, with a mean precision of 96.02% and a mean recall of 95.86%. For cloud shadow detection, the proposed method achieved a mean precision of 95.92% and a mean recall of 92.86%. Experimental results on two validation datasets (GF-1 WFV validation data and ZY-3 validation data) show that the proposed multiscale-3D-CNN method achieved good performance with limited spectral ranges. |
| first_indexed | 2024-12-20T03:51:17Z |
| format | Article |
| id | doaj.art-3088b42e31444fe5bdf64e868104fc26 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-20T03:51:17Z |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-3088b42e31444fe5bdf64e868104fc262022-12-21T19:54:27ZengIEEEIEEE Access2169-35362020-01-018165051651610.1109/ACCESS.2020.29675908962046Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral ImageryYang Chen0https://orcid.org/0000-0002-3407-7845Luliang Tang1https://orcid.org/0000-0003-3523-8994Zihan Kan2https://orcid.org/0000-0002-6364-0537Aamir Latif3https://orcid.org/0000-0001-5435-9885Xiucheng Yang4https://orcid.org/0000-0001-5134-9614Muhammad Bilal5https://orcid.org/0000-0003-1022-3999Qingquan Li6https://orcid.org/0000-0002-2438-6046State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, ChinaInstitute of Geographic Sciences and Natural Resources Research, University of Chinese Academy of Sciences, Beijing, ChinaICube laboratory, University of Strasbourg, Strasbourg, FranceSchool of Marine Sciences, Nanjing University of Information Science and Technology, Nanjing, ChinaState Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, ChinaCloud and cloud shadow detection is one of the most important tasks for optical remote sensing image preprocessing. It is not an easy task due to the variety and complexity of underlying surfaces, such as the low-albedo objects (water and mountain shadow) and the high-albedo objects (snow and ice). In this study, an end-to-end multiscale 3D-CNN method is proposed for cloud and cloud shadow detection in high resolution multispectral imagery. Specifically, a multiscale learning module is designed to extract cloud and cloud shadow contextual information of different levels. In order to make full use of band information, four band-combination images are inputted into the multiscale 3D-CNN. A joint spectral-spatial information of 3D-convolution layer is developed to fully explore the joint spatial-spectral correlations feature in the input data. Overall, in the experiments undertaken in this paper, the proposed method achieved a mean overall accuracy of 97.27% for cloud detection, with a mean precision of 96.02% and a mean recall of 95.86%. For cloud shadow detection, the proposed method achieved a mean precision of 95.92% and a mean recall of 92.86%. Experimental results on two validation datasets (GF-1 WFV validation data and ZY-3 validation data) show that the proposed multiscale-3D-CNN method achieved good performance with limited spectral ranges.https://ieeexplore.ieee.org/document/8962046/Cloud detectioncloud shadowconvolution neural networksmultiscale 3D-CNN |
| spellingShingle | Yang Chen Luliang Tang Zihan Kan Aamir Latif Xiucheng Yang Muhammad Bilal Qingquan Li Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery IEEE Access Cloud detection cloud shadow convolution neural networks multiscale 3D-CNN |
| title | Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery |
| title_full | Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery |
| title_fullStr | Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery |
| title_full_unstemmed | Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery |
| title_short | Cloud and Cloud Shadow Detection Based on Multiscale 3D-CNN for High Resolution Multispectral Imagery |
| title_sort | cloud and cloud shadow detection based on multiscale 3d cnn for high resolution multispectral imagery |
| topic | Cloud detection cloud shadow convolution neural networks multiscale 3D-CNN |
| url | https://ieeexplore.ieee.org/document/8962046/ |
| work_keys_str_mv | AT yangchen cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery AT luliangtang cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery AT zihankan cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery AT aamirlatif cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery AT xiuchengyang cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery AT muhammadbilal cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery AT qingquanli cloudandcloudshadowdetectionbasedonmultiscale3dcnnforhighresolutionmultispectralimagery |