Causality-Driven Infrared and Visible Image Fusion

TL;DR

This paper introduces a causality-based approach to infrared and visible image fusion, addressing dataset bias issues by disentangling causal effects to improve fusion quality.

Contribution

It proposes a novel causal graph framework and a Back-door Adjustment based Feature Fusion Module (BAFFM) to enhance fusion performance by eliminating confounder interference.

Findings

Significantly outperforms state-of-the-art methods on three datasets.

Effectively disentangles causal effects from dataset biases.

Improves robustness and accuracy of image fusion.

Abstract

Image fusion aims to combine complementary information from multiple source images to generate more comprehensive scene representations. Existing methods primarily rely on the stacking and design of network architectures to enhance the fusion performance, often ignoring the impact of dataset scene bias on model training. This oversight leads the model to learn spurious correlations between specific scenes and fusion weights under conventional likelihood estimation framework, thereby limiting fusion performance. To solve the above problems, this paper first re-examines the image fusion task from the causality perspective, and disentangles the model from the impact of bias by constructing a tailored causal graph to clarify the causalities among the variables in image fusion task. Then, the Back-door Adjustment based Feature Fusion Module (BAFFM) is proposed to eliminate confounder interference and enable the model to learn the true causal effect. Finally, Extensive experiments on three standard datasets prove that the proposed method significantly surpasses state-of-the-art methods in infrared and visible image fusion.