MFIF-GAN: A new generative adversarial network for multi-focus image fusion

作者：

Highlights：

• Novel generative adversarial network captures precise boundary features.

• Constructed alpha-matte dataset with defocus spread effect is realistic.

• Efficient and interpretable solution attenuate the defocus spread effect.

• A new diffusion and contraction module confirmed this solution.

摘要

•Novel generative adversarial network captures precise boundary features.•Constructed alpha-matte dataset with defocus spread effect is realistic.•Efficient and interpretable solution attenuate the defocus spread effect.•A new diffusion and contraction module confirmed this solution.

论文关键词：Multi-focus image fusion,Defocus spread effect,Generative adversarial network,Deep learning

论文评审过程：Received 8 November 2020, Revised 4 March 2021, Accepted 11 April 2021, Available online 20 April 2021, Version of Record 23 April 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116295