Video super-resolution based on a spatio-temporal matching network

作者：

Highlights：

• Proposing a simple but effective temporal matching network to assemble video frame features and exploit temporal relations.

• Proposing a novel non-local pyramid residual block, aiming at integrating super-resolution and denoising on wavelet domain.

• Presenting one of the very first attempts towards wavelet domain deep video super-resolution.

摘要

•Proposing a simple but effective temporal matching network to assemble video frame features and exploit temporal relations.•Proposing a novel non-local pyramid residual block, aiming at integrating super-resolution and denoising on wavelet domain.•Presenting one of the very first attempts towards wavelet domain deep video super-resolution.

论文关键词：Deep matching,Wavelet domain,Non-local matching,Residual learning,Video super-resolution

论文评审过程：Received 23 December 2019, Revised 19 August 2020, Accepted 29 August 2020, Available online 2 September 2020, Version of Record 1 November 2020.

论文官网地址：https://doi.org/10.1016/j.patcog.2020.107619