DCFusion: Difference correlation-driven fusion mechanism of infrared and visible images

Min Li, Enguang Zuo, Feng Li, Cheng Chen, Chaoxun Guo, Pei Liu, Yunling Wang, Xiaoyi Lv, Chen Chen

Abstract: Highlights•Implementing multidimensional flexibly collaborative fusion between multimodal image.•Designed loss function of sum of correlation differences to improve interpretability.•Crafted pixel-wise contrast method to adaptively generating salient target masks.•DCFusion achieves SOTA on two dataset in subjective visual and objective evaluation.