Paper Reading Note | Deep Learning for Image Super-resolution: A Survey

一篇深度学习图像超分辨率重建综述，积累一下比较漂亮的句子

Abstract

1 Introduction

With the rapid development of deep learning techniques in recent years, deep learning based SR models have been actively explored and often achieve the state-of-the-art performance on various benchmarks of SR.

2 Problem Setting and Terminology

Although the degradation process is unknown and can be affected by various factors (e.g., compression artifacts, anisotropic degradations, sensor noise and speckle noise),researchers are trying to model the degradation mapping. Most works directly model the degradation as a single downsampling operation, as follows:
压缩假象、各向异性退化、传感器噪声、斑点噪声

高分辨率影像到低分辨率影像通过降采样操作完成，下面涉及机器学习的一些概念了：

2.2 Datasets for Super-resolution

训练数据集的问题

2.3 Image Quality Assessment

重建后的影像质量评估问题

2.3.1 Peak Signal-to-Noise Ratio

峰值信噪比

2.3.2 Structural Similarity

2.3.3 Mean Opinion Score

2.3.4 Learning-based Perceptual Quality

2.3.5 Task-based Evaluation

2.3.6 Other IQA Methods

2.4 Operating Channels

针对哪些波段进行重建的问题

2.5 Super-resolution Challenges

两个挑战，NTIRE Challenge.& PIRM Challenge.

3 SUPERVISED SUPER-RESOLUTION

4 UNSUPERVISED SUPER-RESOLUTION

5 DOMAIN-SPECIFIC APPLICATIONS

6 CONCLUSION AND FUTURE DIRECTIONS

看完再记录……

[1] WANG Z, CHEN J, HOI S C H. Deep Learning for Image Super-Resolution: A Survey [J]. IEEE Trans Pattern Anal Mach Intell, 2021, 43(10): 3365-87.