Infrared-Visible Image Fusion
Overview
Infrared-visible image fusion combines thermal information, which can reveal warm objects in darkness or clutter, with visible-light detail such as edges and textures. Methods include convolutional and Transformer networks, frequency- and wavelet-based fusion, registration for misaligned sensors, and newer diffusion or prompt-guided models that adapt to scene conditions. Researchers assess fused-image quality on benchmarks and test whether fusion improves tasks such as object detection and semantic segmentation. A central challenge is balancing complementary information without introducing artifacts, especially when sensors are misaligned or conditions degrade the images; human-aligned evaluation and reliable gains across tasks remain open concerns.
Latest papers
41 papers6 papers in the last four weeks, up 100% on the four weeks before. 0.0% of all new papers.