FoggyFuse: Infrared and visible image fusion method based on saturation line prior in foggy conditions

Abstract

Infrared and visible image fusion is widely used to enhance image details and information. However, in foggy environments or military smoke bomb scenarios, the scattering and absorption of light significantly degrade the quality of both infrared and visible images, leading to poor fusion performance. Existing fusion methods struggle to effectively restore degraded image details, making them unsuitable for practical applications in such adverse conditions. To address this challenge, we propose a novel fusion architecture based on the saturation line prior (SLP). This method consists of three main modules: the Dehazing Module (DM), the Auxiliary Enhancement Module (AEM), and the Edge Enhancement Module (EEM). The DM optimizes SLP using weighted guided filtering to obtain refined transmission maps for visible images, which are then used to further enhance the infrared image. The AEM and EEM, combined with a non-subsampled shearlet transform (NSST), further process the enhanced visible and infrared images. This approach effectively restores intricate details and achieves natural color reproduction in hazy environments, significantly improving the visual quality of fused images. Given the limited research in this area and the absence of relevant datasets, we constructed an infrared and visible image pair dataset, Foggy, specifically designed for foggy conditions. Qualitative and quantitative evaluations demonstrate that the proposed method outperforms state-of-the-art fusion techniques on the Foggy dataset.