Skip to content
Preprint

When Integral Meets Decomposition: A Signal-Level Self-Supervised Feature Decompose Paradigm for Multi-Modal Image Fusion

Sep 2026 · 0 citations · 51 references
Computer Science

TL;DR

A 1D signal-level self-supervised feature decomposition paradigm to reformulate feature decomposition from unclear 2D image-level supervision into an integral-driven 1D signal-level optimization problem, enabling more stable optimization with a clear optimization objective.

Abstract

Multimodal image fusion (MMIF) aims to integrate complementary information from different modalities into a high-quality fused image and support downstream tasks. Recently, feature decomposition has become an important paradigm by separating source images into common and modality-specific unique features. However, existing methods lack clear supervision because ground-truth (GT) decomposition feature maps are unavailable. They usually combine multiple image-level metrics as losses, which are inherently incomplete and may conflict since each pixel couples attributes such as texture, edge, and contour. To address this, we propose a 1D signal-level self-supervised feature decomposition paradigm. Our core insight is to reformulate feature decomposition from unclear 2D image-level supervision into an integral-driven 1D signal-level optimization problem. This objective-level reformulation uses the 1D signal form to compute the integral constraint. The decomposer is optimized by the integral area between common and original signals, enabling more stable optimization with a clear optimization objective. Our model follows a two-stage SSL framework. Stage I designs dual pretext tasks for integral-driven decomposition at the signal level and structure-preserving reconstruction at the image level. Stage II fuses unique features and combines them with common features to reconstruct the fused image. Experiments on representative MMIF tasks show state-of-the-art (SOTA) performance. Code: github.com/Wangjiayu0512/SIDFusion.

View source

Similar papers

Sep 2026

Beyond Spatial-Domain Supervision: A Relation Constrained Space for Multi-Modal Image Fusion

Multi-modal image fusion (MMIF) aims to form a single image by integrating shared information, preserving complementary cues, and coordinating cross-modal conflicts across modalities. However, due to the absence of ground-truth fused images, existing MMIF supervision commonly uses spatial-domain sources or gradient var...

Ze-Yu Wang, Ming-Yu Ge, Hai-Yu Song et al. · 0 citations
Conference Oct 2026

Infrared and visible image fusion based on nested model decomposition for autonomous driving

To address the problem that visible images degrade severely under nighttime, tunnel, strong backlight, rain and fog conditions, while infrared images can highlight thermal targets but lack detailed texture, this paper proposes an infrared and visible image fusion method based on nested model decomposition. First, laten...

Fei Jiang, Quan-Cheng Du, Yi Shen et al. · 0 citations
Open access Sep 2026

Orthogonal feature decoupling and hierarchical cross-scale aggregation network for RGB-T semantic segmentation

Visible-thermal (RGB-T) imaging systems provide crucial complementary information for robust visual sensing and measurement in complex illumination conditions. However, effectively fusing multi-modal sensory data to achieve accurate semantic segmentation remains a key challenge. Most existing methods rely on heuristic...

Hong-Wei Liu, Yi-Sha Liu, Wei-Min Xue et al. · 0 citations
Sep 2026

LRSP-Fusion: A general image fusion network based on low-rank and sparse priors.

Multi-source images captured from the same scene typically contain both common and unique features, which has been widely exploited in specific image fusion tasks. Existing methods typically consider rough separation of common and unique features for a specific fusion task. However, due to the different modal character...

Shu-Ying Huang, Xi-Wen Xia, Yong Yang et al. · 0 citations
Open access 2026

M2-DINO: Parameter-Efficient and Topology-Oriented Multimodal Fusion for Remote Sensing Image Segmentation

This work proposes M2-DINO, a parameter-efficient multimodal adaptation framework for remote sensing semantic segmentation, where topology-oriented terminology refers to architectural sensitivity to topology-relevant structures, such as thin-object continuity and object boundaries, rather than explicit optimization of...

Zi-Fei Luo, Yu-Bin He, Zhi-Yong Li et al. · 0 citations

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.