Visible and Infrared Image Fusion Using Deep Learning

Area of Science:

• Computer Vision
• Artificial Intelligence
• Machine Learning

Background:

• Visible and Infrared Image Fusion (VIF) is crucial for tasks like object detection and scene segmentation.
• Deep learning methods have rapidly become dominant in VIF over the past five years.
• Existing reviews lack a systematic analysis of these advanced deep learning approaches.

Approach:

• This paper provides a comprehensive review of deep learning-based VIF methods.
• It categorizes VIF techniques, including CNN-based, autoencoder-based, GAN-based, and transformer-based approaches.
• The review details the motivation, taxonomy, development characteristics, datasets, and evaluation metrics for VIF.

Key Points:

• Deep learning has revolutionized VIF, offering superior performance.
• A structured taxonomy of deep learning VIF methods is presented.
• Current datasets and performance evaluation strategies are discussed.

Conclusions:

• This review offers a systematic overview of deep learning-based VIF.
• It serves as a valuable reference for researchers and newcomers to the VIF field.
• Future research directions and prospects in VIF are highlighted.