Face super-resolution via bilayer contextual representation

作者：

Highlights：

• A bilayer contextual representation method for SR is proposed.

• Kernel contextual information has a better ability to learn the complex relationship between LR and HR patches.

• A powerful nonlinear representation scheme is explored to represent rich contextual information from both multilayered structure and kernel function.

摘要

•A bilayer contextual representation method for SR is proposed.•Kernel contextual information has a better ability to learn the complex relationship between LR and HR patches.•A powerful nonlinear representation scheme is explored to represent rich contextual information from both multilayered structure and kernel function.

论文关键词：Face super-resolution,Kernel function,Nonlinear representation,Context-patch

论文评审过程：Received 8 January 2019, Revised 6 March 2019, Accepted 31 March 2019, Available online 11 April 2019, Version of Record 16 April 2019.

论文官网地址：https://doi.org/10.1016/j.image.2019.03.019