A review on infrared and visible image fusion algorithms based on neural networks

Infrared and visible image fusion represents a significant segment within the image fusion domain. The recent surge in image processing hardware advancements, including GPUs, TPUs, and cloud computing platforms, has facilitated the fusion of extensive datasets from multiple sensors. Given the remarkable proficiency of neural networks in image feature extraction and fusion, their application in infrared and visible image fusion has emerged as a prominent research area in recent years. This article begins by providing an overview of the current mainstream algorithms for infrared and visible image fusion based on neural networks, detailing the principles of various image fusion algorithms, their representative works, and their respective advantages and disadvantages. Subsequently, it introduces domain-relevant datasets, evaluation metrics, and some typical application scenarios. Finally, the article conducts qualitative and quantitative evaluations of the fusion results of various state-of-the-art algorithms and offers future research prospects based on experimental results.