Weakly-supervised contrastive learning-based implicit degradation modeling for blind image super-resolution
作者：
Highlights：
• A blind image super-resolution framework is proposed based on degradation modeling.
• We propose to differentiate degradations via weakly-supervised contrastive learning.
• We propose to extract degradation representations via attention-enhanced encoding.
• We develop an effective degradation representation-guided super-resolution network.
• Extensive experiments are conducted to demonstrate the effectiveness of our method.
摘要
•A blind image super-resolution framework is proposed based on degradation modeling.•We propose to differentiate degradations via weakly-supervised contrastive learning.•We propose to extract degradation representations via attention-enhanced encoding.•We develop an effective degradation representation-guided super-resolution network.•Extensive experiments are conducted to demonstrate the effectiveness of our method.
论文关键词：Blind image super-resolution,Implicit degradation modeling,Weakly-supervised contrastive learning,Attention mechanism,Deep convolutional neural networks
论文评审过程：Received 6 January 2022, Revised 22 April 2022, Accepted 3 May 2022, Available online 11 May 2022, Version of Record 21 May 2022.
论文官网地址：https://doi.org/10.1016/j.knosys.2022.108984
