SlimDiffSR: Toward Efficient Remote Sensing Image Super-Resolution via Diffusion Model Distillation

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