How to define a circle in an image and calculate the percentage it is filled.

I'm looking to calculate the percentage of each dotted circles that is filled by the red spots within the attached image. Also attached is a code I have previously ceated that calculates the size of the red spots by filtering the image and using region props, as well as the size of the dotted circle by defining its diameter. However, this only provides a size comparison between the two circles regardless of their location. I want to alter the code to find the percentage of the black dotted circle that is covered by the red. My idea was to define the area of the dotted circles and determine the area of each regionprop within that circle, but I am struggling to do so. Any guidence would be greatly appreciated.

Thank You!

Answer 1

[Simon Chan]

Some modification from your code as follows:

(1) Use your code to detect the red circles

(2) Generate convex hull image to calculate the centroid of each red circle

(3) Find the dotted circles by thresholding

(4) Separate the entire image into 3 parts

(5) Find the dots belongs to the dotted circle and fit a circle

(6) Generate a mask from the fitted circle

(7) Union red circle mask and the dotted circle mask, gives your required percentage.

clc; close all;
img = imread('https://www.mathworks.com/matlabcentral/answers/uploaded_files/899000/circlesize.jpg');
[R,G,B] = imsplit(img);
gray_img = rgb2gray(img);
xz_red = imsubtract(R,gray_img);
xz_red = imadjust(xz_red,stretchlim(xz_red),[]);
bw = im2bw(xz_red);
%%  Use your code to detect the red circles above
bw1 = bwconvhull(bw,'object');                      % Generate convex hull image
s1 = regionprops(bw1);                              
centroid = cat(1,s1.Centroid);                      % Find the centroid of each red circle
%%  Find the dotted circles
red_range = [60,120];
mask1 = R>red_range(1) & R<red_range(2);
bw11 = imclearborder(mask1);                        % Clear border
bw12 = bwareaopen(bw11,50);                         % Remove small objects
%
[Ny,Nx,Nc] = size(img);
x_coord = [1;round(centroid(1:2,1)+diff(centroid(:,1))/2);Nx];      % Separate into 3 pictures
t = tiledlayout(2,2,'TileSpacing','compact','Padding','compact');
%
for k = 1:length(x_coord)-1
    xstart = x_coord(k);                            % Start x-coordinates
    xend = x_coord(k+1);                            % End x-coordinates
    regionA = bw1(:,xstart:xend);                   % Extract the red circle
    regionB = bw12(:,xstart:xend);                  % Extract the dotted circle
    nexttile
    imshow(img(:,xstart:xend,:),[]);                % Showing original image
    hold on;
    [row,col] = find(bwperim(regionA));             % For display purpose only
    plot(col,row,'g.','MarkerSize',3);              % Display the peimeter of red circle
    %% Fit a circle for the dotted circle
    [row,col] = find(regionB);                      % Find all white dots belongs to the dotted circle
    row = row(:);
    col = col(:);
    a = [row col ones(size(row))]\(-(row.^2+col.^2));
    yCircle = -.5*a(1);                             % y-center of fitted circle
    xCircle = -.5*a(2);                             % x-center of fitted circle
    rCircle = sqrt((a(1)^2+a(2)^2)/4-a(3));         % radius of fitted circle
    %
    [Nyroi,Nxroi] = size(regionB);
    [xx,yy] = meshgrid(1:Nxroi,1:Nyroi);
    mask = false(Nyroi,Nxroi);
    mask = mask | hypot(xx - xCircle, yy - yCircle) <= rCircle;     % Creating a mask for the fitted circle
    insideA = sum(regionA & mask)/sum(mask);                     % Calculate the percentage
    rectangle('Position',[xCircle-rCircle yCircle-rCircle, 2*rCircle, 2*rCircle],'Curvature',[1 1]);    %Display only
    title(sprintf('Image #%d, Percentage = %.1f %%',k,100*insideA));
    hold off;
end

[Image Analyst]

You can use viscircles() instead of rectangle() to display circles.