Improving the Fusion Effect of Multiple Elements in Visual Communication by Image Fusion Algorithm

Image fusion technology plays an increasingly important role in the field of visual communication, but under extreme conditions, including underexposure, overexposure, and severe noise, there are significant problems with the image fusion effect. In addition, images captured by different types of sensors often have different resolutions. How to overcome the resolution difference and integrate the effective information from different source images is a challenge. In order to solve these problems, this paper uses image fusion algorithm to make improvements. First, the illumination-aware progressive image fusion network PIAFusion estimates the lighting conditions through a dedicated sub-network and calculates the illumination probability, thereby constructing the illumination-aware loss. Next, the scene illumination decoupling network SIDNet eliminates degraded illumination features, and the color consistency loss is dedicated to reducing color distortion in the fused image while injecting more visible domain information into the fused image. Finally, the mutually enhanced multimodal image registration and fusion MURF treats image registration and fusion as a whole rather than an isolated task through the shared information extraction module SIEM, the multi-scale coarse registration module MCRM, and the fine registration and fusion module F2M. The results show that under extreme lighting conditions, the image fusion algorithm used in this paper can improve the multi-element fusion effect under extreme conditions, and in terms of color consistency, the CIE 2000 color difference value is the lowest, and the NSS value in the naturalness evaluation is the highest, reaching 0.96, an advantage in color consistency and naturalness. In addition, there are also significant advantages in cross-resolution image fusion effects, and it is effective in improving image visual quality and information transmission efficiency.