Search code examples
imagematlabcolorsobject-detectionmask

MATLAB - Changing color of a portion of an image by applying a mask

I've got this coloured image and I obtained the following mask by doing object detection. I would like to apply this mask to the original image to change the color of the solids to the one that I like (or apply any effect to them, it doesn't really matter). I'm having trouble doing that.

I've tried creating a function that iterates over every pixel of the original image comparing it to the mask but it's not an efficient solution.

The mask is of logical type while the image is uint8

enter image description hereenter image description here

Solution

  • Ugly images below, but hopefully also an instructive example:

    % Create dummy image
x = 0:4:255;

origimage = uint8(repmat(x,64,1));
origimage(:,:,2) = 100;
origimage(:,:,3) = 100;

% create dummy mask
colormask = logical(zeros(64,64));
colormask(1:20,30:50) = 1;

% display original image and mask

figure(1);image(origimage);
figure(2); imagesc(colormask);


% create altered image

newimage = origimage;
newimage(colormask) = 255; %changes first channel to 255 in masked pixels
figure(3); image(newimage);

    Example of altering image in a mask region

    Since colormask is a logical, you can use it directly to select the relevant pixels in the original image and modify them in one command. Hope this helps.

    **Added 7/12/2023 in response to comment about changing other channels:

    You have flexibility to change the other channels. My original code only changed the red channel because my mask was only 2D.

    % change all three channels together

colormask2 = logical(zeros(64,64,3));
colormask2(1:20,30:50,1:3) = 1;

newimage2 = origimage;
newimage2(colormask2) = 200;

    Alternate and more flexible approach:

    %create 3D mask with different numbers for each channel
colormask3 = uint8(zeros(64,64,3));
colormask3(1:20,30:50,1) = 1;
colormask3(1:20,30:50,2) = 2;
colormask3(1:20,30:50,3) = 3;
        
newimage4 = origimage;
newimage4(logical(colormask3)) = 200; %gives same image as prior example
figure(4); image(newimage4);
        
newimage5 = origimage;
newimage5 = newimage5 + 10*colormask3; %adds 10 to red, 20 to green, 30 to blue
figure(5); image(newimage5);
        
%  add to the original image itself *1/2 for red, *1 for green, and *1.5
%  for blue
newimage6 = origimage + origimage.*colormask3/2;
figure(6); image(newimage6);