Detail-preservation and saliency-extraction-based infrared and visible image fusion

Infrared and visible image fusion can combine detail information of visible image and saliency features of visible image, resulting in images that are not only more suitable for human visual perception but also enhance performance in various machine tasks. Despite these advantages, most fusion methods ignore tiny textures and insufficiently preserve saliency target information. To overcome these challenges, we propose a novel infrared and visible image fusion method, which can sufficiently retain weak detail information and adjust the whole image illumination by a detail enhancement-based fusion strategy and adaptive weights. Our method utilize pyramid to decompose the source image into base layer, detail layer and saliency layer, which can mitigate the loss of fine-textured details during the fusion process. Additionally, We enhanced the details during the fusion process employing a window-based fusion method that maximizes the retention of intricate texture information. Qualitative and quantitative experimental results conducted on public datasets demonstrate the superiority and effectiveness of our proposed method.