Change color from a circle form in an image

Hello. I have to find circles in an image and to change their colors. I found the circles, but I don't know how to detect the color and change.

Answer 1

[Luca Ferro]

try to take a look here (https://www.mathworks.com/matlabcentral/answers/20391-finding-a-specific-colour-within-a-picture#answer_26933) for color detection.

And here (https://www.mathworks.com/matlabcentral/answers/11256-changing-colors-in-an-image) to change colors.

[DGM]

Regarding the latter topic, this answer also applies and includes links to several other demos on various means to change colors.

Answer 2

[Image Analyst]

The color is the RGB values of the image array. Not sure what you mean by detect, but if you want a mask of certain colors, then try using the Color Thresholder on the Apps tab of the tool ribbon. Try HSV color space and click the button to export code.

See my Image Processing Tutorial in my File Exchange:

Image Analyst's File Exchange

Answer 3

[DGM]

I'm just going to throw this out there as one idea. You could also pick colors in a region near the known center (if you can rely on the center being clear).

This uses MIMT tools, but the links include examples that don't use MIMT. The core idea here is just a means to find the colored regions by identifying the most common color within each circle. This isn't really a good approach generally, but for this case with broad regions of uniform color.

% so you have an RGB image
inpict = imread('circlecircle.png');
inpict = im2double(inpict); % this is needed later

% say you already know where the circles are regardless of their color
C = [68 74; 134 115];
R = [51 58];

% create an indexed copy to simplify color picking
[indpict map] = rgb2ind(inpict,64);

% preallocate mask
ncircles = numel(R);
mask = false([size(indpict) ncircles]);

% sample those locations based on the circle geometry
imshow(inpict); % set up axes to use ROI tools
for b = 1:ncircles
    % create mask
    ROI = images.roi.Circle(gca);
    ROI.Center = C(b,:);
    ROI.Radius = R(b);
    circmask = createMask(ROI);
    
    % sample region in indexed copy
    samplepx = indpict(circmask);
    
    % find color of most common index
    % integer-class indexed images index from zero
    modecolor = map(mode(samplepx)+1,:)
    modecolor = permute(modecolor,[1 3 2]);
    
    % find all pixels in the ROI close to the most common color in the ROI
    tol = 0.1; % pick some tolerance
    thismask = all(abs(inpict-modecolor) <= tol,3) & circmask;
    mask(:,:,b) = imclose(thismask,ones(3)); % fill in thin lines (optional)
end
% now you have masks for the colored region within each circle or partial circle
% you can use those in whatever way you choose

% alter the hue in each region
hueshift = rand(2,1); % amount to shift hue for each blob (random example)
outpict = inpict;
for b = 1:ncircles
    % these are both from MIMT, but there are non-MIMT examples
    modpict = imtweak(outpict,'hsl',[hueshift(b) 1 1]);
    outpict = replacepixels(modpict,outpict,mask(:,:,b));
end

imshow(outpict)

[Image Analyst]

One might also play around with lazysnapping