Discover MakerZone

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

Learn more

Discover what MATLAB® can do for your career.

Opportunities for recent engineering grads.

Apply Today

Can you help me to correct this error?

Asked by Tomas on 23 Nov 2012

Hi, can you help me to correct this error

[p C]=size(lab)
Lr = L .* ones(p, C);


??? Error using ==> times
Matrix dimensions must agree.
Error in ==> lab at 60
Lr = L .* ones(p, C);






No products are associated with this question.

3 Answers

Answer by Matt Fig on 23 Nov 2012
Edited by Matt Fig on 23 Nov 2012

What are p and C?

L = 60;
Lr = L .* ones(3, 5);  % No error...(the .* is not necessary)

You have a variable named lab and a function named lab?? This is a very bad idea and may (somehow) be at the root of your problem....


Tomas on 23 Nov 2012
>> whos p C ones
  Name      Size            Bytes  Class     Attributes
    C         1x1                 8  double              
    p         1x1                 8  double              
p =
C =
Matt Fig on 23 Nov 2012

There is more to the story then. See, this does not error on my machine:

p = 400;
C = 1800;
L = 60.0221;
Lr = L .* ones(p, C);

So I am afraid I am out of ideas here. There is something else going on in the code somewhere, but without seeing the code I cannot say. Someone else may be able to guess the problem.

Walter Roberson on 24 Nov 2012

Tomas, what are the values of p and C ? And is the error still occurring?

Matt Fig
Answer by Walter Roberson on 24 Nov 2012

Your RGB image is 3 dimensional, not 2 dimensional. Row, Column, Color-plane. Your lab equivalent of the RB is 3 dimensional as well. When you take the size() of a 3 dimensional array, you need to have either only 1 output or else 3 or more outputs. If you only have two outputs as in

[p C]=size(lab)

then p will be assigned the number of rows (as expected) but C will be assigned the product of all the remaining dimensions, and so will be columns times number of color planes.

1 Comment

Matt Fig on 24 Nov 2012

That doesn't explain the error, however...

>> clear all
>> rgb=imread('office_2.jpg');    % Image in the IPT
>> cform = makecform('srgb2lab'); % On the FEX
>> lab = applycform(rgb,cform);
>> [p C]=size(lab);
>> % lchan=lab(:,:,1);
>> L=60.0221;
>> Lr = L .* ones(p, C);   % No error....
>> whos
  Name         Size                  Bytes  Class     Attributes
    C            1x1                       8  double              
    L            1x1                       8  double              
    Lr         600x2709             13003200  double              
    cform        1x1                   13376  struct              
    lab        600x903x3             1625400  uint8               
    p            1x1                       8  double              
    rgb        600x903x3             1625400  uint8 
Walter Roberson
Answer by Image Analyst on 24 Nov 2012
Edited by Image Analyst on 24 Nov 2012

Like Walter said, your problem is (partially) that your C is the number of columns TIMES the number of color channels (which is 3). You should be doing it this way, as you'll notice that I ALWAYS do it in code I post, even when I suspect the image is gray level:

[rows columns numberOfColorChannels] = size(lab);

If you do it this way:

[rows columns] = size(lab);

as you did, then columns will not be the number of columns in the image but will be columns * numberOfColorChannels. Please see this blog entry by Steve Eddins for further discussion:

But like Matt said, that wouldn't cause an error if L was just some number like 60. So, I suspect that you actually did something like

L = lab(:, :, 1); % Extract the L color channel.

and then L would have the size rows*columns, but your ones() function would return an array rows*columns*numberOfColorChannels. Hence the sizes don't match up. So I don't know why you said L was 60 when it was not, but that's your problem. I speculate that you'd like to construct a constant luminance image. In that case, you set

Lr = 60 * ones(size(L));

and then you invert the transform to get back to RGB.


Image Analyst

Contact us