A bias-free Bayesian framework for generalizable and robust multi-modality image fusion
Authors:
- Shihan MaSMShihan Ma
- Hong SongHSHong Song
- Jinfu LiJLJinfu Li
- Ting WangTWTing Wang
- Yucong LinYLYucong Lin
- Jianghan XiaJXJianghan Xia
- Yuanyuan WangYWYuanyuan Wang
- Tianyu FuTFTianyu Fu
- Jingfan FanJFJingfan Fan
- Danni AiDADanni Ai
- Deqiang XiaoDXDeqiang Xiao
- Jian YangJYJian Yang
Preprint
- 30 July 2026
- preprint
- Published by Elsevier in SSRN Electronic Journal
Abstract
Multi-Modality Image Fusion (MMIF) aims to reconstruct an informative latent image from complementary sensor observations and is commonly formulated as an inverse problem within a Bayesian framework. Due to its ill-posed nature, the recovery process critically depends on prior knowledge to constrain the solution space and preserve underlying structural characteristics. However, current methods primarily rely on hand-crafted, task-specific priors, which not only introduce subjective biases but also undermine generalizability and robustness across diverse real-world scenarios. To address this issue, we reformulate MMIF within a bias-free Bayesian framework. Each modality is modeled as an independent degradation of the fused image, transforming conventional predefined priors into cross-modality conditional priors and enabling a fully data-driven fusion process. Building on this foundation, we propose the Symmetric Degradation Complementarity Unrolling Network (SDCUNet), which explicitly unrolls the optimization process into a symmetric and interpretable architecture. Each stage of SDCUNet comprises a Degradation Awareness Module (DAM) to model the modality-specific degradation, and a Complementarity Enhancement Module (CEM) to extract structural guidance from cross-modality conditional priors. Additionally, we introduce a Local Gradient-Guided (LGG) pixel-level loss function to adaptively quantify each modality's contribution, further enhancing the model's robustness against complex environmental degradations. Extensive experiments on five infrared-visible and two medical image datasets demonstrate that SDCUNet consistently achieves superior visual quality and quantitative metrics, exhibiting remarkable robustness even under challenging degradation conditions, while improving the downstream detection performance. The code is available at https://github.com/op-dog/SDCUNet.Keywords
This publication has 28 references indexed in Scilit:
- DDBFusion: An unified image decomposition and fusion framework based on dual decomposition and Bézier curvesInformation Fusion, 2024
- DIVA: Deep unfolded network from quantum interactive patches for image restorationPattern Recognition, 2024
- A general image fusion framework using multi-task semi-supervised learningInformation Fusion, 2024
- FATFusion: A functional–anatomical transformer for medical image fusionInformation Processing & Management, 2024
- MUFusion: A general unsupervised image fusion network based on memory unitInformation Fusion, 2022
- PIAFusion: A progressive infrared and visible image fusion network based on illumination awareInformation Fusion, 2022
- Image fusion in the loop of high-level vision tasks: A semantic-aware real-time infrared and visible image fusion networkInformation Fusion, 2022
- Infrared and visible image fusion via gradient transfer and total variation minimizationInformation Fusion, 2016
- A general framework for image fusion based on multi-scale transform and sparse representationInformation Fusion, 2015
- Convergence of descent methods for semi-algebraic and tame problems: proximal algorithms, forward–backward splitting, and regularized Gauss–Seidel methodsMathematical Programming, 2011