MDFN: Mask deep fusion network for visible and infrared image fusion without reference ground-truth

作者：

Highlights：

• A deep fusion network is proposed for fusing visible and infrared images.

• Weight score strategy estimates contributions of two source images in pixel level.

• Mask strategy keeps the fused image consistent with two source images.

• A hybrid loss function is designed to preserve the visible and thermal information.

摘要

•A deep fusion network is proposed for fusing visible and infrared images.•Weight score strategy estimates contributions of two source images in pixel level.•Mask strategy keeps the fused image consistent with two source images.•A hybrid loss function is designed to preserve the visible and thermal information.

论文关键词：Image fusion,Mask strategy,Deep learning,Weight score estimation,Visible and infrared images

论文评审过程：Received 25 January 2022, Revised 24 June 2022, Accepted 17 August 2022, Available online 28 August 2022, Version of Record 5 September 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.118631