Current Trends in Infrared-Visible Image Fusion (IVIF) Techniques, Evaluation Metrices and Open Issues

Abstract

 Infrared-visible image fusion (IVIF) is a major effort in computer vision meant to enhance the representation of pictures by combining the unique qualities of the infrared and visible wavelengths. Though they can run into problems like feature redundancy, complexity, and computational inefficiency, conventional fusion methods rely on hand-crafted algorithms to find and combine important features. Recent developments in deep learning have significantly improved fusion performance by means of adaptive feature extraction and the elimination of computing overhead. Still, there is a clear lack of a thorough assessment of DL-based fusion techniques and their impact on the quality and efficiency of picture fusion.  Emphasizing deep learning approaches, this study presents a thorough investigation of IVIF methods. This paper addresses major issues including data compatibility, perceptual accuracy, and evaluation constraints while exploring a variety of fusion methods and assessing their advantages and limitations. While identifying potential paths for research—such as simplified network design and hardware-efficient fusion techniques, the report underlines the merits and shortcomings of present methods. Establishing a basis for future development in IVIF technology, this work is a vital tool for academics and professionals.