Conti-Fuse: A novel continuous decomposition-based fusion framework for infrared and visible images

Hui Li, Haolong Ma, Chunyang Cheng, Zhongwei Shen, Xiaoning Song, Xiaojun Wu

Abstract: Highlights•Non-overlapping sparse decomposition hampers the preservation of key information.•Decomposed features of source images act as samples in a deep feature space.•Monte Carlo adds randomness to the loss, improving model robustness.•Overlapping dense decomposition aids fusion, boosting downstream task performance.