Image Fusion in Remote Sensing: An Overview and Meta-Analysis

Remote sensing image fusion is consistently used to turn raw images of different resolutions, sources, and modalities into accurate, complete, and spatiotemporally coherent images. It facilitates downstream applications such as pan sharpening, change detection, and classification. However, image fusion solutions are highly disparate to various remote sensing problems and are often narrowly defined in existing reviews as topical applications (e. g., pan sharpening). Theoretically, image fusion can be applied to any gridded data through pixel-level operations; thus, we expand its scope by comprehensively surveying relevant works. We develop a simple taxonomy for many-to-one and many-to-many image fusion, defining it as a mapping problem turning one or multiple images into another set based on desired coherence. Furthermore, we provide a meta-analysis to cover 10,420 peer-reviewed papers from the 1980s to 2023 studying various types of image fusion and their applications. Finally, we discuss image fusion's benefits and emerging challenges to provide open research directions.