GrFormer: A novel Transformer on Grassmann manifold for infrared and visible image fusion

Huan Kang, Hui Li, Xiaojun Wu, Tianyang Xu, Rui Wang, Chunyang Cheng, Josef Kittler

Abstract: Highlights•Grassmann subspaces extend spatial/channel attention for multi-modal fusion.•Low-rank mappings disentangle image semantics for multi-scale cross-modal learning.•Cross-modal strategy enhances foreground-background information separation.•Our manifold-based fusion approach shows superior benchmark performance.

External IDs:dblp:journals/inffus/KangLWXWCK26