Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

关于重构的公式内的傅里叶变换 #38

Open
sixgod696 opened this issue May 31, 2023 · 4 comments
Open

关于重构的公式内的傅里叶变换 #38

sixgod696 opened this issue May 31, 2023 · 4 comments

Comments

@sixgod696
Copy link

想请问作者,内核重构公式内和dcls最后的H里的傅里叶变换是主要是起什么作用?低通滤波吗?

@sixgod696
Copy link
Author

我了解到的低通滤波主要是为了压缩高频元素,来减弱下采样的混叠。作者你调换了顺序,采用傅里叶变换,是为了达到类似的效果吗?这里没有太明白,希望作者可以解惑

@Algolzw
Copy link
Collaborator

Algolzw commented Jun 4, 2023

想请问作者,内核重构公式内和dcls最后的H里的傅里叶变换是主要是起什么作用?低通滤波吗?

因为在频域里面处理去卷积(或去模糊)比较方便,这个跟其它去模糊论文类似。最简单的例子就是空域的卷积就是频域的乘积,而去模糊过程就只需要在频域里使用模糊图像除以卷积核就可以得到清晰图像。

@Algolzw
Copy link
Collaborator

Algolzw commented Jun 4, 2023

我了解到的低通滤波主要是为了压缩高频元素,来减弱下采样的混叠。作者你调换了顺序,采用傅里叶变换,是为了达到类似的效果吗?这里没有太明白,希望作者可以解惑

我们调换顺序是因为超分里的输入是低分辨率空间,而原始的模糊核是在高分辨率空间(HR原图)上,我们采用核变换后将高分辨率的模糊核等价变到了低分辨率空间里,这样就可以用传统的去卷积/去模糊手段等得到干净图像再进行普通超分。也就是说解耦了去模糊和超分两个过程。

@sixgod696
Copy link
Author

我基本明白了,非常感谢大佬解惑。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants