3d to 2d matrix

Hi guys,

I have an [r c w] matrix where r & c are the pixels of a hyperspectral picture and the w are the number of layers of the picture.

I want to turn this matrix to [w , x] matrix.

For example. I want the values of w for the pixel 1,1 below 1. and the values of w of the pixel 1,2 below 2 until the whole table is finished.

I thought of the following commands but I'm not sure if it works because my matrix is massive.

[r,c,w]=size(reflectances) A=reshape(reflectances,w,r*c);

Thank you.

Answer 1

[Azzi Abdelmalek]

 [r,c,w]=size(reflectances)  
 A=reshape(reflectances(:),r*c,[])'

[Björn]

In order to make this work, you should change the first line into:

 [r,c,w]=size(reflectances)

[Azzi Abdelmalek]

Thanks Björn, when I changed my code, I missed it

Answer 2

[Björn]

This might do the trick:

 A=reshape(reflectances,1,r*c,w);
 A=reshape(A,r*c,w)';

You first convert it to a [1,r*c,w] matrix. After this you convert it to a [r*c,w] matrix and taking the transpose resulting in an [w,r*c] array with the values you wanted.

Answer 3

[Andrei Bobrov]

A = reshape(reflectances.',[],w).';

[Björn]

Taking the transpose of an array with more than 2 dimensions does not work

Answer 4

[Alexandros Polykarpou]

IS there anyway to check if the final form of the matrix is what I want. The matrix is quite big... so any ideas? ans =

      192000         356

[Björn]

Just check a small portion of the initial matrix and the final matrix:

 reflectances(1:5,1:5,1:5)
 A(1:25,1:5)

Compare these answers with each other. You can also try to check it with a dummy-matrix. Create a small matrix with random values, and then check the final answer. I did that with this code:

 test_matrix=randn(4,5,3)

Then apply the method you want to try (replace 'reflectances' in the code by 'test_matrix')

And then you can easily check if the first values of each third dimension coincides with the values of the first column, and so on.

NOTE: I also fixed a typo in my first answer