DiffFuSR: Super-Resolution of all Sentinel-2 Multispectral Bands using Diffusion Models

2025-06-13

AI Summary

To address the inconsistent spatial resolutions (10/20/60 m) across the 12 spectral bands of Sentinel-2 Level-2A imagery, this work proposes the first end-to-end blind super-resolution framework to uniformly enhance all bands to a 2.5 m ground sampling distance (GSD). The method adopts a two-stage modular design: first, a diffusion model super-resolves the RGB bands; second, a generative-prior-driven multispectral fusion architecture leverages the super-resolved RGB guidance to upsample the remaining bands. Key innovations include blind degradation modeling and a contrastive degradation encoder, integrated with multi-scale feature extraction and cross-domain harmonized training on NAIP and WorldStrat datasets. Evaluated on the OpenSR benchmark, our approach comprehensively surpasses state-of-the-art methods—particularly improving spatial detail and spectral fidelity in the 20 m and 60 m bands—outperforming conventional pan-sharpening techniques.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

This paper presents DiffFuSR, a modular pipeline for super-resolving all 12 spectral bands of Sentinel-2 Level-2A imagery to a unified ground sampling distance (GSD) of 2.5 meters. The pipeline comprises two stages: (i) a diffusion-based super-resolution (SR) model trained on high-resolution RGB imagery from the NAIP and WorldStrat datasets, harmonized to simulate Sentinel-2 characteristics; and (ii) a learned fusion network that upscales the remaining multispectral bands using the super-resolved RGB image as a spatial prior. We introduce a robust degradation model and contrastive degradation encoder to support blind SR. Extensive evaluations of the proposed SR pipeline on the OpenSR benchmark demonstrate that the proposed method outperforms current SOTA baselines in terms of reflectance fidelity, spectral consistency, spatial alignment, and hallucination suppression. Furthermore, the fusion network significantly outperforms classical pansharpening approaches, enabling accurate enhancement of Sentinel-2's 20 m and 60 m bands. This study underscores the power of harmonized learning with generative priors and fusion strategies to create a modular framework for Sentinel-2 SR. Our code and models can be found at https://github.com/NorskRegnesentral/DiffFuSR.

Problem

• Super-resolving all Sentinel-2 bands to 2.5m GSD
• Blind SR using diffusion models and fusion networks
• Outperforming SOTA in fidelity and spectral consistency

Innovation

• Diffusion-based super-resolution for Sentinel-2 RGB
• Fusion network upscales multispectral bands
• Contrastive degradation encoder supports blind SR