RADiffSR: A Diffusion Model for Remote Sensing Image Super-Resolution Fusing Residual Attention and Cross-Scale Dynamic Gating
Bibliographic record
Abstract
Remote sensing image super-resolution (SR) technology is critical for enhancing the fine interpretation capability of large-scale land cover elements. However, existing methods are constrained by three core deficiencies: insufficient information interaction in multispectral channel modeling, lack of spatiotemporal continuity modeling for geographic entities, and failure of cross-scale feature geometric alignment. These deficiencies lead to coupled challenges in reconstructed images, including morphological discontinuities of extensive geographic features and texture artifact proliferation. This paper proposes a remote sensing image SR algorithm based on the diffusion probabilistic model (DPM), referred to as RADiffSR. First, a Residual-Attention Enhancement Block (RAE Block) is designed. It integrates residuals and Nonlinear Activation-Free Block to form a dual-domain attention mechanism, which synchronously optimizes feature response weights in the spatial and spectral domains, alleviating the deficiency in correlation representation between multispectral channels. Second, we introduce large-kernel convolutional layers to construct multi-level receptive field architectures aligned with geographic entity scale characteristics, modeling extensive terrain continuity through enlarged kernel sizes while incorporating inverted bottleneck ConvFFN structures to deepen feature extraction and implicitly enhance high-frequency texture retention. Finally, a feature manifold alignment strategy is implemented with dynamic gating mechanisms between encoder-decoder pathways to regulate cross-scale feature propagation weights, suppressing semantic distortion and high-frequency information loss. We construct the GF7-SR super-resolution dataset based on GF-7 satellite imagery, encompassing diverse typical land cover scenarios including mountainous houses, farmland, forests, and water bodies for model training and testing. Experiments demonstrate that RADiffSR achieves 36.39 dB and 28.36 dB PSNR on GF7-SR and Toronto datasets respectively, significantly outperforming state-of-the-art methods.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".