An efficient method of infrared-visible image fusion

Fusing visible and infrared images enhances the situational awareness of optoelectronic detection systems in complex environments. In this study, we propose a fast and robust algorithm for infrared-visible image fusion. Specifically, we decompose infrared and visible images into detail and base layers using structured decomposition, and then guide the base layer fusion using weight parameter maps obtained through visual saliency detection (VSD). Finally, we fuse the detail and base layer based on the entropy maximization rule to produce the final fused image. Our method enriches image data, retaining and highlighting detailed image information. This approach establishes a strong foundation for downstream high-level visual tasks, such as target recognition and tracking. Experimental results show that compared with other fusion methods, our proposed method has excellent competitiveness on different data sets. Real-time performance of 1024×1280 infrared and visible images with a frame rate of nearly 50Hz is achieved on an embedded platform.