SRCNN阅读笔记

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

1. prediction models,

2. edge based methods,

3. image statistical methods

4. 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