Shift-Invariant Unsupervised Pansharpening Based on Diffusion Model
Open Access
- 22 December 2025
- journal article
- research article
- Published by MDPI AG in Remote Sensing
- Vol. 18 (1) , 27
Abstract
Pansharpening is a crucial topic in remote sensing, and numerous deep learning-based methods have recently been proposed to explore the potential of deep neural networks (DNNs). However, existing approaches are often sensitive to spatial translation errors between high-resolution panchromatic (HRPan) and low-resolution multispectral (LRMS) images, leading to noticeable artifacts in the fused results. To address this issue, we propose an unsupervised pansharpening method that is robust to translation misalignment between HRPan and LRMS inputs. The proposed framework integrates a shift-invariant module to estimate subpixel spatial offsets and a diffusion-based generative model to progressively enhance spatial and spectral details. Moreover, a multi-scale detail injection module is designed to guide the diffusion process with fine-grained structural information. In addition, a carefully formulated loss function is established to preserve the fidelity of fusion results and facilitate the estimation of translation errors. Experiments conducted on the GaoFen-2, GaoFen-1, and WorldView-2 datasets demonstrate that the proposed method achieves superior fusion quality compared with state-of-the-art approaches and effectively suppresses artifacts caused by translation errors.Keywords
This publication has 55 references indexed in Scilit:
- CSLP: A novel pansharpening method based on compressed sensing and L-PNNInformation Fusion, 2025
- Stable classification with limited samples in global land cover mapping: Theory and experimentsScience Bulletin, 2024
- Unsupervised hyperspectral pansharpening via low-rank diffusion modelInformation Fusion, 2024
- A review of regional and Global scale Land Use/Land Cover (LULC) mapping products generated from satellite remote sensingISPRS Journal of Photogrammetry and Remote Sensing, 2023
- Diffusion model with disentangled modulations for sharpening multispectral and hyperspectral imagesInformation Fusion, 2023
- Zero-shot semi-supervised learning for pansharpeningInformation Fusion, 2023
- Zero-Sharpen: A universal pansharpening method across satellites for reducing scale-variance gap via zero-shot variationInformation Fusion, 2023
- Modified N-FINDR endmember extraction algorithm for remote-sensing imageryInternational Journal of Remote Sensing, 2015
- A Variational Model for P+XS Image FusionInternational Journal of Computer Vision, 2006
- A wavelet transform method to merge Landsat TM and SPOT panchromatic dataInternational Journal of Remote Sensing, 1998