Set to zero largest wavelet coefficients.

1 次查看（过去 30 天）
Gabriel Melendez
Gabriel Melendez 2020-3-10
Hi everybody,
I am reading the book "A Mathematical Introduction to Compressive Sensing".
The next example took my attention.
I cant believe it! So I am trying to repeat the above example.
There is not more information about the type of wavelets used to modify the 99% of the coeffcients.
So I computed 'haar DWT' coefficients of the cameraman image as next:
image = imread('cameraman.tif');
wavename = 'haar';
[cA,cH,cV,cD] = dwt2(im2double(image),wavename);
Then, I set to zero high frequency sub-bands coefficients and apply the inverse DWT.
cH(:) = 0;
cV(:) = 0;
cD(:) = 0;
mod_image = idwt2(cA,cH,cV,cD,wavename);
I only modified the 75% of the total coefficients, however the recovered image loses a lot of quality (Compared with the example, where 99% of the coefficients are set to zero).
What should I do to repeat the example?
Should I use a different wavelet type?
How should I modify the largest coefficients?
Thank you very much for your help!

请先登录，再回答此问题。

回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Discrete Multiresolution Analysis 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by