How to change a matrix consisting of 3 elements into another matrix of 3 different elements?

I have a 47x47 matrix. The matrix entries consist of the values 0, 96, and 255. All entries with the value of 255 I want to transform to the value of 0. All entries with the value of 96 I want to transform to the value of 3. All entries with the value of 0 I want to transform to the value of 1. Does anyone how I could do this easily?

Answer 1

[Ameer Hamza]

Try this

A; % 47x47 matrix

A_new = zeros(size(A));
A_new(A==255) = 0;
A_new(A==96) = 3;
A_new(A==0) = 1;

Read about logical indexing: https://www.mathworks.com/company/newsletters/articles/matrix-indexing-in-matlab.html

[Borys Bulka]

Hey! Thank you for your answer but It tells me "Error: Incorrect use of '=' opperation. To assign a value to a variable, use '='. To compare values for equality, use '=='.

[Borys Bulka]

Oh but it does work when I write it as a code and not in the command window!

[Borys Bulka]

However, now that I have this matrix A and have another matrix B (also 47x47). It doesnt let me create a matrix C=A+B. It says: 'Intergers can only be combined with integers of the same class, or scalar doubles.'

[Ameer Hamza]

Can you show your code? Can you confirm if A and B have equal sizes?

[Borys Bulka]

I did confirm that they are both equal size with the size function. I will copy and past the code in a second.

[Borys Bulka]

A=imread('cm_young.png');

B=imresize(A, [611 611]);

C=[B([7:13:605],[7:13:605])];

am_young=zeros(size(C));

am_young(C==255)=0;

am_young(C==96)=0.5;

am_young(C==0)=1;

D=imread('cm_chung.png');

E=imresize(D, [517 517]);

am_chung=[E([6:11:512],[6:11:512])];

[r,c]=size(am_chung);

for i=1:r

for j=1:c

if am_chung(i,j)<180

am_chung(i,j)=1;

else

am_chung(i,j)=0

end

end

end

am_new=am_young+am_chung

[Borys Bulka]

Everything works as it supposed to work, expect I get an error at an_new. It doesn't calculate am_new for me.

[Borys Bulka]

I changed the 3 to a 0.5 by the way, since I see it must be a 0.5 instead of a 3.

[Borys Bulka]

It says that am_young is a class: double and am_chung is a class: uint8. Is it because of that? I am not very experienced with Matlab so I don't know all those things that are mentioned there.

[Ameer Hamza]

Ok. I get it. You are mixing two things here. By default, MATLAB loads images in unit8 format in which the pixel intensities vary from 0 to 255. It appears that you want the image intensities in range 0 to 1. Although there are several ways to handle this situation, However, I recommend to do everything by considering that pixel intensity vary from 0 to 255

A=imread('cm_young.png');
B=imresize(A, [611 611]);
C=[B([7:13:605],[7:13:605])];
am_young=zeros(size(C));
am_young(C==255)=0;
am_young(C==96)=0.5*255;
am_young(C==0)=1*255;
am_young = uint8(am_young);

D=imread('cm_chung.png');
E=imresize(D, [517 517]);
am_chung=[E([6:11:512],[6:11:512])];
[r,c]=size(am_chung);
for i=1:r
    for j=1:c
        if am_chung(i,j)<180
            am_chung(i,j)=255;
        else
            am_chung(i,j)=0
        end
    end
end
am_new=am_young+am_chung

[Borys Bulka]

it does calculate the am_new matrix then but the values then aren't as they supposed to be. Becuase now the values 255 and 128 are in the matrix am_new. While the am_new matrix should only contain values 0, 0.5, and 1 after adding the am_chung matrix to the am_young matrix.

[Ameer Hamza]

how are they different?

[Borys Bulka]

When doing 0.5*255, the value becomes 128, but this value must remain 0.5. When doing 1*255, the value becomes 255, but this value must remain 1.

[Borys Bulka]

When i read the images first and used different forms of thresholds and or transformation I could normally just add the two matrices up. But when using this kind of transformation:

am_young=zeros(size(C));

am_young(C==255)=0;

am_young(C==96)=0.5*255;

am_young(C==0)=1*255;

am_young = uint8(am_young);

It somehow changed the class of the matrix such that I could not add them up anymore.

[Ameer Hamza]

What does the value of 1 represent in your code? Does it mean maximum brightness of pixel? For a unit8 image, value of 1 is equivalent to 255. If you look at the values in matrix 'C'. You will see they do not vary from 0 to 1, they vary from 0 to 255.

[Borys Bulka]

cm_young is an image of a black grey white coloured 47x47 connectivity matrix with black, grey and white squares. I read this image in matlab and resized it such that I could take each middle pixel of each square in the 47x47 connectivity matrix image. a black pixel has in matlab a value of 0 and thus I wanted to transform every 0 to a 1. every grey pixel has a value of 96, so I want to transform that to a 0.5. Every white pixel has a value of 255, so I wanted to transform that to a value of 0. Doing this I should receive a adjacency matrix that was used to construct the original image (am_young).

So this matrix:

A=imread('cm_young.png');

B=imresize(A, [611 611]);

C=[B([7:13:605],[7:13:605])];

am_young=zeros(size(C));

am_young(C==255)=0;

am_young(C==96)=0.5*255;

am_young(C==0)=1*255;

am_young = uint8(am_young);

does not give me the same output when using:

A=imread('cm_young.png');

B=imresize(A, [611 611]);

C=[B([7:13:605],[7:13:605])];

am_young=zeros(size(C));

am_young(C==255)=0;

am_young(C==96)=0.5;

am_young(C==0)=1;

[Ameer Hamza]

Oh! If you wanted to do that then you can directly use im2double()

A=imread('cm_young.png');
B=imresize(A, [611 611]);
C=[B([7:13:605],[7:13:605])];
C=im2double(C);

it will automatically scale the value between 0 and 1. Similarly, you can do the same for

am_chung=[E([6:11:512],[6:11:512])];
am_chung=im2double(am_chung);

However, it will linearly map values between 0 to 1. If you must want 96 at 0.5 then try the following code

A=imread('cm_young.png');
B=imresize(A, [611 611]);
C=[B([7:13:605],[7:13:605])];
am_young=zeros(size(C));
am_young(C==255)=0;
am_young(C==96)=0.5;
am_young(C==0)=1;

D=imread('cm_chung.png');
E=imresize(D, [517 517]);
am_chung=[E([6:11:512],[6:11:512])];
am_chung_ = zeros(size(am_chung));
[r,c]=size(am_chung);
for i=1:r
    for j=1:c
        if am_chung(i,j)<180
            am_chung_(i,j)=1;
        else
            am_chung_(i,j)=0
        end
    end
end

am_new=am_young+am_chung_

[Borys Bulka]

I used the last code you wrote and it work! Thank you very much for your patients and help I appreciate it a lot! I am however a bit frustrated with that I do not understand why this works and my first attempt did not work. If you'd have the will still to be willing to explain that in baby steps I would highly appreciate it since I'd like to understand. Or perhaps know a link that would explain that better for me? If not, this is perfectly fine! :)

[Borys Bulka]

So I can the only difference was to add the zeros(size()) part. Why did that change it so that I then could add them up and beforehand I couldn't?

[Ameer Hamza]

Here is a brief explanation.

These lines

A=imread('cm_young.png');
B=imresize(A, [611 611]);
C=[B([7:13:605],[7:13:605])];

load image as uint8 matrix, in which the pixel value change from 0 to 255. Note that uint8 can only save integer values, i.e., you cannot save 0.5. But then you run the line

am_young=zeros(size(C));

By default, zeros() function contain 'double' datatype matrix, i.e., it can save fractional values, therefore, following lines work properly

am_young(C==255)=0;
am_young(C==96)=0.5;
am_young(C==0)=1;

Then again you load the image

D=imread('cm_chung.png');
E=imresize(D, [517 517]);
am_chung=[E([6:11:512],[6:11:512])];

and it again create uint8 matrix. But when you run

am_chung_ = zeros(size(am_chung));

it create am_chung_ with double datatype. am_chung_ and am_young both have double datatype, therefore, the code works properly.

[Borys Bulka]

Thank you very much for your time and effort! I appreciate it alot, it makes much more sense to me know!