Year: 2024
Type: article
Abstract: . The problem of cloud cover significantly affects the quality and utility of optical satellite imagery, posing challenges for image analysis and interpretation. In response, this research presents a fast and efficient image reconstruction method by exploiting the strength of isophote in linear structure propagation of ground features, providing users with a practical option for cloud removal task. In addition, this research solves the inapplicability problem when the cloud-contaminated regions overlap between different temporal images, and then extends the application to the field of multi-temporal data cloud removal. Through several experiments, the proposed method outperforms other comparative methods for both dual-temporal data reconstruction and multi-temporal data reconstruction task. Furthermore, the method proves to be efficient and robust even with limited auxiliary information. The proposed multi-temporal image reconstruction framework is available at: https://github.com/YuXiaoyu221/ICMIR.
Authors: Xiaoyu Yu, Jun Pan, Jiangong Xu, Junli Li
Language: English
Subfield: Ocean Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
Awards: 高分六号宽视场影像大区域全覆盖有效成图研究, G7891852649
OpenAlex ID: W4404162638