<p>With the increasing complexity of power systems, monitoring electrical equipment has become increasingly important. Traditional monitoring methods rely on a single type of image data, making it difficult to comprehensively capture the status of equipment. To address this, this paper proposes an improved infrared and visible light image fusion network based on salient object detection, named STDFusionNet+. We have added a Transformer module to the original STD network and introduced an Aggregated Residual Dense Block (ARDB), then applied it to the power equipment infrared and visible light image registration dataset. This network efficiently retains thermal targets in infrared images and texture structures in visible light images. By using a salient object template to label regions of interest in the infrared images and designing a specific loss function, the network can selectively extract salient object features from the infrared images and background texture features from the visible light images, ultimately generating efficient fusion results. We conducted experiments with this network on our power equipment infrared and visible light image registration dataset. A large number of qualitative and quantitative experimental results demonstrate the superiority of our method, showing significant advantages in six metrics: EN, MI, VIF, SF, SSIM and PSNR.</p>