In global Earth observation, multispectral imaging is frequently hindered by extensive cloud cover, leading to observation gaps. Introducing synthetic aperture radar (SAR) data with all-weather penetration capability for multimodal cloud removal has become a key approach to achieving global spatiotemporal seamless remote sensing monitoring. Existing SAR-optical fusion methods often lack effective decoupling and differentiated processing of heterogeneous data. This leads to texture distortion in reconstructed images, limiting the accuracy of downstream interpretation and target recognition. Therefore, a cloud removal method based on SAR-guided alignment and multifrequency collaborative enhancement is proposed in this article. First, the spatial geometric features of SAR images are mined by multireceptive field gating mechanism, and robust structural priors are extracted for under-cloud ground object reconstruction while effectively suppressing coherent speckle noise; Second, with the help of the deformable alignment module, the geometric alignment of SAR and optical images is realized with reference to optical images, alleviating the problem of scale and spatial misalignment; Finally, through the multifrequency collaborative enhancement module, the high and low frequency information are adaptively separated, and an improved attention mechanism is adopted to enhance the high and low frequency information, respectively, which effectively suppresses cloud interference and maintains surface details. Results on the M3R-CR and LuojiaSET-OSFCR datasets show that the proposed method comprehensively outperforms the other eight compared methods. Compared with the suboptimal method, the peak signal-to-noise ratios (PSNRs) achieved by the proposed method on the two datasets are improved by 0.5643 and 0.1964 dB, respectively. The source code of SAMCE-CR is shared at https://github.com/RSIDEA-ECUT/SAMCE-CR
Shu-Ting Yang, Xun-Qiang Gong, Xiu-Fang Zhou et al.· IEEE Transactions on Geoscie...· 0 citations
Remote sensing open-vocabulary scene classification aims to recognize unseen scene categories beyond the predefined training set by leveraging external knowledge and vision-language models. Prompt learning has emerged as an effective paradigm to adapt pretrained vision-language models to open-vocabulary remote sensing image processing tasks by dynamically encoding task-specific knowledge into textual prompts. However, most existing approaches mainly rely on category names derived from training data, which lack fine-grained semantic descriptions and attribute-level associations that are crucial for complex remote sensing scenes. This limitation often degrades generalization when encountering novel classes and diverse geospatial domains. To address this problem, we propose KnowProKD, a knowledge-guided framework that incorporates diverse external knowledge and constraint-based regularization to enhance base-to-novel generalization. Specifically, KnowProKD consists of three components: 1) leveraging large language models (LLMs) to generate diverse attribute and description knowledge for constructing knowledge-aware prompts, improving cross-modal semantic alignment; 2) introducing knowledge-based constraints to regularize training and enhance robustness to novel category recognition; and 3) designing a dual-branch logits calibration mechanism that exploits attribute and description prompts to balance performance across base and novel classes. Extensive experiments on eight challenging remote sensing scene classification benchmarks demonstrate that KnowProKD achieves favorable performance compared with recent prompt-learning baselines and generalizes well to previously unseen scenarios. Furthermore, we apply KnowProKD to cross-regional urban land-use mapping and statistical analysis across representative cities from six continents, validating its practical effectiveness in real-world remote sensing applications.
A. Ma, Weihao Shen, Rui-Yi Yang et al.· IEEE Transactions on Geoscie...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
