# GSPan: A Continuous Gaussian Primitive Representation for Arbitrary-Scale Pansharpening

arXiv: https://arxiv.org/abs/2606.17722 · HTML: https://mlsanity.com/paper/2606.17722/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 48 references that are in the map.
- First posted on arXiv 16 Jun 2026.

## Builds on (most cited first, 19 of 19)

- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [3D Gaussian Splatting for Real-Time Radiance Field Rendering](https://mlsanity.com/paper/2308.04079/) (2023)
- [Learning Continuous Image Representation with Local Implicit Image Function](https://mlsanity.com/paper/2012.09161/) (2020)
- [A Survey on 3D Gaussian Splatting](https://mlsanity.com/paper/2401.03890/) (2024)
- [Local Texture Estimator for Implicit Representation Function](https://mlsanity.com/paper/2111.08918/) (2021)
- [GaussianImage: 1000 FPS Image Representation and Compression by 2D Gaussian Splatting](https://mlsanity.com/paper/2403.08551/) (2024)
- [Deep Learning in Multimodal Remote Sensing Data Fusion: A Comprehensive Review](https://mlsanity.com/paper/2205.01380/) (2022)
- [Pan-Mamba: Effective pan-sharpening with State Space Model](https://mlsanity.com/paper/2402.12192/) (2024)
- [CiaoSR: Continuous Implicit Attention-in-Attention Network for Arbitrary-Scale Image Super-Resolution](https://mlsanity.com/paper/2212.04362/) (2022)
- [Pixel to Gaussian: Ultra-Fast Continuous Super-Resolution with 2D Gaussian Modeling](https://mlsanity.com/paper/2503.06617/) (2025)
- [Remote Sensing Image Fusion Based on Two-stream Fusion Network](https://mlsanity.com/paper/1711.02549/) (2017)
- [PSGAN: A Generative Adversarial Network for Remote Sensing Image Pan-Sharpening](https://mlsanity.com/paper/1805.03371/) (2018)
- [Generalized and Efficient 2D Gaussian Splatting for Arbitrary-scale Super-Resolution](https://mlsanity.com/paper/2501.06838/) (2025)
- [Target-adaptive CNN-based pansharpening](https://mlsanity.com/paper/1709.06054/) (2017)
- [Unsupervised Hyperspectral Pansharpening via Low-rank Diffusion Model](https://mlsanity.com/paper/2305.10925/) (2023)
- [PanFormer: a Transformer Based Model for Pan-sharpening](https://mlsanity.com/paper/2203.02916/) (2022)
- [Hyperspectral and multispectral image fusion with arbitrary resolution through self-supervised representations](https://mlsanity.com/paper/2405.17818/) (2024)
- [Multi-Head Attention Residual Unfolded Network for Model-Based Pansharpening](https://mlsanity.com/paper/2409.02675/) (2024)
- [G-ZAP: A Generalizable Zero-Shot Framework for Arbitrary-Scale Pansharpening](https://mlsanity.com/paper/2603.14412/) (2026)

## Abstract (arXiv)

Pansharpening aims to generate high-resolution multispectral (HRMS) images by fusing low-resolution multispectral (LRMS) and panchromatic (PAN) observations. Most existing deep learning methods treat pansharpening as fixed-grid prediction, which limits scale adaptation. To address this, we propose GSPan, a framework that introduces 2D Gaussian Splatting (GS) into pansharpening. Instead of directly predicting pixels, GSPan represents band-wise residual details as continuous and learnable 2D Gaussian primitives. We design a Dual-Stream Hierarchical Interaction (DSHI) architecture with a Spatial-Spectral Interactive Attention (SSIA) module to estimate these primitives from complementary PAN and MS observations. The predicted primitives are rendered as a residual detail field and injected into the upsampled MS image. This continuous representation allows GSPan to render fused images on arbitrary target sampling grids without scale-specific retraining. It further enables a Scale-Decoupled Asymmetric Inference (SDAI) strategy, which estimates primitives at a reduced resolution and renders the fused image at the target resolution for efficient large-scene pansharpening. Experiments on QuickBird, GaoFen-2, WorldView-3, and WorldView-3-4K datasets show that GSPan delivers state-of-the-art fusion performance. Moreover, SDAI markedly accelerates inference, achieving a favorable trade-off between computational efficiency and fusion quality. Our results demonstrate the potential of continuous Gaussian residual representations as a flexible and scale-decoupled alternative to fixed-grid prediction.