标题: Generative Shadow Synthesis and Removal for Remote Sensing Images Through Embedding Illumination Models
作者: Shao, CL (Shao, Chenglin); Li, HF (Li, Huifang); Shen, HF (Shen, Huanfeng)
来源出版物: IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 卷: 63 文献号: 5620115 DOI: 10.1109/TGRS.2025.3561307 Published Date: 2025
摘要: Shadows significantly reduce the available information in remote sensing images, obstructing downstream tasks such as object detection, scene classification, and localization. Shadow removal from remote sensing images is, however, still an open issue for the following reasons. First, deep neural networks are difficult to train since the corresponding ground truth of shadows is almost always unavailable in practice. Second, the existing shadow removal methods still suffer from blurry details and boundary artifacts. In this article, we describe how a generative shadow synthesis and removal framework that couples data-driven methods with illumination models was developed to address the above challenges effectively. Various shadows were synthesized in shadow-free regions of remote sensing images by GSS-Net, which is a generative shadow synthesis network that considers the physical process of shadow illumination attenuation (SIA). In this way, a large-scale, diverse, and realistic shadow dataset (RS-SynShadow) was built. A generative shadow removal network-GSR-Net-embedding a histogram-enhanced illumination (HEI) model was then developed for high-fidelity shadow removal without artifacts. Extensive experiments conducted on synthetic and real data demonstrate that the proposed shadow synthesis and removal framework significantly outperforms the state-of-the-art methods, both visually and quantitatively.
作者关键词: Remote sensing; Lighting; Deep learning; Training; Image color analysis; Attenuation; Visualization; Generative adversarial networks; Data models; Data mining; Generative learning; illumination models; remote sensing images; shadow removal; shadow synthesis
KeyWords Plus: AERIAL IMAGES
地址: [Shao, Chenglin; Li, Huifang] Wuhan Univ, Sch Resource & Environm Sci, Wuhan 430079, Peoples R China.
[Shen, Huanfeng] Wuhan Univ, Sch Resource & Environm Sci, Key Lab Geog Informat Syst, Minist Educ, Wuhan 430079, Peoples R China.
[Shen, Huanfeng] Wuhan Univ, Key Lab Digital Cartog & Land Informat Applicat, Minist Nat Resources, Wuhan 430079, Peoples R China.
通讯作者地址: Li, HF (通讯作者)，Wuhan Univ, Sch Resource & Environm Sci, Wuhan 430079, Peoples R China.
电子邮件地址: [email protected]; [email protected]; [email protected]
影响因子：7.5
