Abstract: Infrared and visible image fusion aims to integrate complementary information from two modalities to enhance visual perception under adverse conditions, and has been widely applied in night vision, autonomous driving, and surveillance. However, most existing methods assume clean inputs and ignore the presence of complex degradations, such as haze, rain, and low illumination, leading to poor robustness in real-world applications. To address this issue, we propose a novel deep fusion framework named DAME-Fusion, which incorporates a degradation-aware mixture-of-experts (MoE) module and a modality reconstruction mechanism. Specifically, the MoE module adaptively selects enhancement experts based on degradation types (e.g., fog, rain, and low-light), while the reconstruction branch enhances cross-modal consistency. Extensive experiments conducted on four public datasets (LLVIP, MSRS, M3FD, and RoadScene) show that our method achieves state-of-the-art performance across multiple metrics, including average gradient (AG), entropy (EN), spatial frequency (SF), standard deviation (SD), visual information fidelity (VIF), and mutual information (MI). Compared with existing approaches, our method produces fusion results with clearer detail, better structure preservation, and more consistent semantic alignment, validating its effectiveness and practical potential in degraded environments.