SRCNN阅读笔记
Abstract
on-line
INTRODUCTION
1、we show that the aforementioned pipeline
is equivalent to a deep convolutional neural network [26]
证明
2、 Our method differs fundamentally from existing
external example-based approaches, in that ours does not
explicitly learn the dictionaries [39], [47], [48] or manifolds
[2], [4] for modeling the patch space.
manifold 是什么?
3、The proposed SRCNN has several appealing properties.
- simplicity ,superior accuracy
- our method achieves fast speed for practical on-line usage
- the restoration quality of the network can be further improved
cope with three channels of color images simultaneously to
achieve improved super-resolution performance.
4、对之前版本的改进
- improve the SRCNN by introducing larger filter size in the non-linear mapping layer, and explore deeper structures by adding non-linear mapping layers.
- process three color channels (either in YCbCr or RGB color space) simultaneously.
- considerable new analyses and intuitive explanations are added to the initial results.
测试集增加了BSD200,变为Set5, Set14,BSD200
2 RELATED WORK
2.1 Image Super-Resolution
single-image super resolution algorithms:分为四类
-
prediction models,
-
edge based methods,
-
image statistical methods
-
patch based (or example-based) methods
achieve the state-of-the-art performance.(1) The internal example-based methods 原理 (2) The external example-based methods 原理
基于内部实例方法和外部实例方法的区别:
These studies vary on how to learn a compact
dictionary or manifold space to relate low/high-resolution
patches, and on how representation schemes can be con-
ducted in such spaces.
通道的处理:
- The majority of SR algorithms [2], [4], [15], [39], [46], [47],[48], [49] focus on gray-scale or single-channel image super-resolution.
- . For color images, the aforementioned methods
first transform the problem to a different color space (YCbCr or YUV), and SR is applied only on the luminance channel. - There are also works attempting to super-resolve all chan-
nels simultaneously.
none of of them has analyzed the SR performance of different channels, and the necessity of recovering all three channels.
2.2