A question about different image size generation using Matlab

Hello everyone!

I need your help! Can anyone please tell me why, the Y1 becomes empty and its size changes when I input a true color image ?

X1 =  double(imread('true_color_image'));
Y1 = padarray(X1(end-1:end,end-1:end),[1 1],'replicate','pre')
size(X1)
size(Y1)

Y1 =

       0     0     0
       0     0     0
       0     0     0

ans =

     120   120     3

ans =

       3     3

While when I use input data, as shown below, Y1 does not change:

X2 =  [1 2 3; 4 5 6; 7 8 9]
Y2 = padarray(X2(end-1:end,end-1:end),[1 1],'replicate','pre')
size(X2)
size(Y2)

X2=

       1     2     3
       4     5     6
       7     8     9

Y2=

       5     5     6
       5     5     6
       8     8     9

ans =

       3     3

ans =

       3     3

What can I do so that the size of Y1 does not change (i.e. remains same as X1) ?

Thank you!

[Image Analyst]

If X1 is really a true color image like the filename suggests, then why are you only indexing it with only 2 indexes instead of 3? Do you realize what that actually does when you do that (I'd guess not)?

[Gobert]

What you have just suggested does not lead to any answers to my question.@ Image Analyst

X1 =  double(imread('true_color_image'));
Y1 = padarray(X1(end-1:end,end-1:end,1:end),[1 1],'replicate','pre');
size(X1)
size(Y1)

ans =

     120   120     3

ans =

       3     3     3

[Image Analyst]

What does this say

[rows, columns, numberOfColorChannels] = size(X1)

[Gobert]

@ Image Analyst . It simply says this:

X1 =  double(imread('true_color_image'));
[rows, columns, numberOfColorChannels] = size(X1)

rows =

     120

columns =

     120

numberOfColorChannels =

       3

[Image Analyst]

Looks like you figured it out now since you've accepted an answer.

Answer 1

[Guillaume]

X1(end-1:end,end-1:end)

Since you're using 2D indexing to index a 3D matrix, this only return values from the 1st plane (the red colour plane). This is equivalent to

X1(end-1:end,end-1:end, 1)

If you want a 3D matrix as output, you must pass a 3D matrix as input, so:

Y1 = padarray(X1(end-1:end,end-1:end, :), [1 1], 'replicate', 'pre')

Note that the above will work on greyscale images as wel, since the : will match the only colour plane.

[Gobert]

@ Guillaume , even if I consider the 3rd dimension, I still don't get the expected Y1 size - and, that is my question, I would like to know why or what I can do to get it fixed: See this example with your correction:

X1 =  double(imread('true_color_image'));
Y1 = padarray(X1(end-1:end,end-1:end, :), [1 1], 'replicate', 'pre');
size(X1)
size(Y1)

ans =

     120   120     3

ans =

       3     3     3

[Guillaume]

Well, end-1:end only select the last 2 columns/rows. So padding it by 1 is always going to give you 3 columns/rows.

Matlab is doing exactly what you're asking. What you're asking is obviously not what you really mean. You haven't say what it is though.

If you want to pad the whole array:

Y1 = padarray(X1, [1 1], 'replicate', 'pre');

[Gobert]

Thank you very much, Guillaume .

I got it:

X1 =  double(imread('true_color_image'));
Y1 = padarray(X1(2:end,2:end,:), [1 1], 'replicate', 'pre');
size(X1)
size(Y1)

ans =

     120   120     3

ans =

     120   120     3