2026· IEEE Transactions on Geoscience and Remote Sensing· Vol 64, pp. 5640718-5640718· 0 citations· 74 references
TL;DR
A structured pruning strategy tailored to remote sensing imagery is presented, which systematically removes redundant semantic modules and replaces standard operations with lightweight designs, including frequency-separable convolution, direction-separable convolution, and a query-driven global aggregation module (QGAM).
Abstract
Diffusion models (DMs) have recently achieved remarkable performance in image super-resolution (SR), but their high computational cost limits practical deployment in remote sensing applications. To address this issue, we propose SlimDiffSR, an efficient diffusion-based framework for real-world remote sensing image SR. Unlike existing single-step diffusion methods that rely on fixed timesteps, we first introduce an uncertainty-guided timestep assignment strategy to construct a stronger single-step teacher model, where reconstruction difficulty is explicitly linked to diffusion timesteps, enabling adaptive generative strength. Building upon this teacher, we further present a structured pruning strategy tailored to remote sensing imagery, which systematically removes redundant semantic modules and replaces standard operations with lightweight designs, including frequency-separable convolution (FSC), direction-separable convolution (DSC), and a query-driven global aggregation module (QGAM). These components explicitly exploit the unique characteristics of remote sensing data, such as sparse high-frequency details, strong directional patterns, and long-range spatial dependencies. To strengthen the distillation process, we introduce maximum mean discrepancy (MMD)-based distribution alignment alongside mean squared error (mse) feature supervision, improving peak signal-to-noise ratio (PSNR) by 0.23 dB. Extensive experiments on multiple remote sensing benchmarks demonstrate that SlimDiffSR achieves a favorable balance between efficiency and reconstruction quality. In particular, it attains up to $200\times $ inference acceleration and a $20\times $ reduction in model parameters compared with multistep DMs, while achieving competitive perceptual quality and clearly outperforming existing efficient diffusion baselines in efficiency. The code is available at https://github.com/wwangcece/SlimDiffSR
Remote sensing image super-resolution (SR) fundamentally relies on effective feature representation learning. However, the existing methods that adopt mainstream architecture designs, such as transformer and state space model (SSM), are often restricted by two factors when dealing with the SR of remote sensing images....
Kang-Hui Zhao, Tao Lu, Jia-Ming Wang et al.· IEEE Transactions on Geoscie...· 0 citations
Image super-resolution (SR) aims to recover a high-resolution (HR) image from its low-resolution (LR) counterpart, yet accurately reconstructing texture details in complex regions remains challenging. This issue is particularly pronounced in remote sensing images (RSIs), which contain diverse textures and small-scale o...
Cui-Cui Lv, Jia-Jie Wang, Shu-Zhen Xu et al.· IEEE Transactions on Geoscie...· 0 citations
Although self-attention has advanced remote sensing image superresolution (RSISR), its high computational cost and large-scale variations in remote sensing imagery hinder efficient deployment on edge devices. Given this, we propose a lightweight RSISR network, called the direction-aware residual network (DireNet), whic...
Wei Xue, Meng-Cheng Ma, Ming-Yang Du et al.· IEEE Geoscience and Remote S...· 0 citations
Layout-controlled remote sensing image generation is valuable for data augmentation, scene simulation, and object-level geospatial understanding. However, it remains challenging due to complex backgrounds, dense small objects, rotated objects, and complex object-context relationships. To address the limitations of exis...
A detection-oriented super-resolution framework built upon a residual diffusion-based super-resolution branch and guided by downstream detection supervision is developed, which improves mAP50 from 0.5655 for low resolution inputs to 0.5899, while maintaining competitive reconstruction fidelity.
Ming-Ming Xie, Hong-Wei Li, Meng-Ni Xu et al.· ITU Journal on Future and Ev...· 0 citations
Remote sensing image super-resolution aims to reconstruct high-quality spatial details from low-resolution observations, which is essential for land-cover interpretation, object detection, and downstream geospatial analysis. However, existing convolutional neural network (CNN)- and Transformer-based methods may struggl...
Shuai Liu, J. Fonseca, André Mora· Remote Sensing· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.