LAUSR

Infrared images can rely on the thermal radiation of objects for imaging, independent of lighting conditions. Furthermore, because the thermal radiation produced by targets such as people, vehicles, and boats differs greatly from the background, it is able to distinguish objects from their environment as well. These characteristics of infrared can be complemented with visible images, which are rich in color information but vulnerable to lighting conditions. Therefore, the fusion of IR and visible images can provide a better perception of the environment. In this paper, we propose a new infrared–visible fusion algorithm. It consists of three parts: feature extraction, fusion, and reconstruction. The attention mechanism is introduced into the feature extraction to better extract features and we propose a new way of describing the fusion task. The relationship between the two inputs is balanced by introducing a fused image obtained by summing the infrared and visible images. It is also optimized for sky layering and water surface ripples, which are common in water environments. The edge information is enhanced in the loss function and noise reduction is performed. Through comparison experiments, our algorithm achieves better results.