Finding, accessing and replacing specific pixel values in image?

I am working on a task and need help in syntax for finding specific pixel values,then accessing them and finally replace them.

I have implemented this in a long iterative fashion. But it is very cumbersome and if I need to make some changes in the assigned colors I have to make changes in a lot of places.

'I' is the input image a 527x527 logical indexed image. G is the true color output image of the same size that I am creating. Its code is the first block.

'totalLabels' is a vector of size 8,each value being unique, e.g. [1 4 14 24 56 77 110 120]. These are the labels that I have assigned to my input image 'I' and then I replace each of them with a color of my choosing in the new RGB image 'G'.

I also sort of brain stormed a little :-), and implemented a shorter version using 'find' and 'switch' commands. That is second block of code. But it gives an error.

G = zeros(w,l,3);

%Much longer iterative version
for rr = 1:w
    for cc = 1:l
        px = I(rr,cc);
        if px ~= 0
          if px==totalLabels(1)
              G(rr,cc,1) = 1.0;
              G(rr,cc,2) = 0;
              G(rr,cc,3) = 0;
          end
          if px==totalLabels(2)
              G(rr,cc,1) = 0;
              G(rr,cc,2) = 1.0;
              G(rr,cc,3) = 0;
          end
        if px==totalLabels(3)
              G(rr,cc,1) = 0;
              G(rr,cc,2) = 0;
              G(rr,cc,3) = 1.0;
        end
        if px==totalLabels(4)
              G(rr,cc,1) = 0.2;
              G(rr,cc,2) = 0.8;
              G(rr,cc,3) = 0.1;
        end

           if px==totalLabels(5)
                G(rr,cc,1) = 0.3;
                G(rr,cc,2) = 0.5;
                G(rr,cc,3) = 0.1;
           end
           if px==totalLabels(6)
                G(rr,cc,1) = 0.9;
                G(rr,cc,2) = 0.8;
                G(rr,cc,3) = 0.4;
           end

            if px==totalLabels(7)
                G(rr,cc,1) = 0.4;
                G(rr,cc,2) = 0.2;
                G(rr,cc,3) = 0.1;
            end

            if px==totalLabels(8)
                G(rr,cc,1) = 0.6;
                G(rr,cc,2) = 0.6;
                G(rr,cc,3) = 0.1;
            end
          end
      end
  end

figure;imshow(G);

%shorter crisp code with errors
G = zeros(w,l,3);
for xx = 1:numel(totalLabels)
    [r,c] = find(I==totalLabels(xx));
    switch xx
        case 1
            G(r,c) = 20;
        case 2
            G(r,c) = 55;
        case 3
            G(r,c) = 85;
        case 4
            G(r,c) = 100;
        case 5
            G(r,c) = 125;
        case 6
            G(r,c) = 140;
        case 7
            G(r,c) = 160;
        case 8
            G(r,c) = 180;
        otherwise
            G(r,c) = 0;
    end
end

My question is this: Is there a crisper,shorter way of doing what I am trying to do?

Although the iterative approach works but it is very long. I have attached the Input labelled logical image 'I' and 'totalLabels' vector for any one interested in trying. ('w' and 'l' are found by [w,l,~] = size(I);)

Answer 1

[Brattv]

Hi, try this code

% Loading data
G = load('image_plus_labels.mat'); 

% Image 
image = G.I;
% Labels 
totalLabels = G.totalLabels;

% Making a cell of colors 
colors = {[1 0 0], [0 1 0], [0 0 1], [0.2 0.8 0.1], [0.3 0.5 0.1],...
    [0.9 0.8 0.4], [0.4 0.2 0.1], [0.6 0.6 0.1]}

% Size of image 
[w l] = size(image);

% Making the zero matrix 
G = zeros(w,l,3);

for i=1:length(totalLabels)
    % Finding the index of labels
    indexOfInterest = find(image == totalLabels(i));

      % Offset to access all dimensions.
      indexOffset = 0;
      for j=1:3
          G(indexOfInterest+indexOffset) = colors{i}(j);
          indexOffset = indexOffset + 277729;
      end 

end 

imshow(G)

Notice that you can change the 277729 integer in indexOffset with the total amount of pixels in the image. I think

indexOffset = indexOffset + length(image(:))

might be a working alternative. Hope it gives you the result you were looking for.

[Hassan Javaid]

This generates a 3D image as output. Like real 3D with depth perception. I am looking for a planar RGB image.

[Hassan Javaid]

Okay I tweaked the code. And got the result that I needed. And in good time also i.e. 0.138424 seconds. Your syntax for cell arrays helped.

% Making a cell of colors 
colors = {[1 0 0], [0 1 0], [0 0 1], [0.2 0.8 0.1], [0.3 0.5 0.1],...
    [0.9 0.8 0.4], [0.4 0.2 0.1], [0.6 0.6 0.1]};
for i=1:numel(totalLabels)
    % Finding the index of labels
    [rows,cols] = find(I == totalLabels(i));
      % Iterating through indices to assign color.
      len = numel(rows);
      for j=1:len
          G(rows(j),cols(j),:) = colors{i};
      end 
end