1D Haar wavelet transform

how to Decompose query image using Haar Wavelet

transformation at 1st level

then I want to Calculate mean and standard deviation of each coefficient and form one dimensional feature vector.

I don't know how to use wavelet Toolbox

Answer 1

[Wayne King]

If you are sure you just want the 1st-level (and no further), you can use dwt2.

If you need multiple levels, use wavedec2().

However, your statement " mean and standard deviation of each coefficient and form one dimensional feature vector." does not make sense.

Do mean the mean and standard deviation of each subband image?

You can easily obtain the subband images

    load woman;
    dwtmode('per');
    [CA,CH,CV,CD] = dwt2(X,'haar');

CH, CV, and CD are the wavelet subband images. Now can you point me to some text which describes how to find your feature vector from those matrices?

[shima said]

Decompose query image using Haar Wavelet transformation at 1st level to get approximate coefficient and vertical, horizontal and diagonal detail coefficients.

After a one-level wavelet transform, the wavelet coefficients is ci, j at the point (i, j), then the mean and standard deviation of any band are calculated as:μ =1/

[Wayne King]

The above is incomplete: "then the mean and standard deviation of any band are calculated as:μ =1/ "

[shima said]

I am sorry I sent it incomplete y mistake

After a one-level wavelet transform, the wavelet coefficients is ci, j at the point (i, j), then the mean and standard deviation of any band are calculated as:μ =1/

[Wayne King]

It is still not complete. What you posted is no different than your previous incomplete post.

[shima said]

I am very sorry I don't konw how did that happe :( i copied it and when i sent it looked like that

[Walter Roberson]

shima commented,

I want to containing the mean value and the variance of each image's coefficients.