Try this code in order to find the center of your "circle"
% some constants for "circle" detection
GRAY_TH = 128; % gray threshold
ECCENTRICITY_TH = 0.5; % eccentricity threshold, from 0(=>circle) to 1(=>line)
% load and convert to a binary image
entropy_img = imread('entropyfilt_imgB.jpg');
bw_entropy_img = entropy_img < GRAY_TH;
imshow(bw_entropy_img); hold on;
% look for region properties
s = regionprops(bw_entropy_img, entropy_img, {'Eccentricity', 'Centroid'});
% keep and plot those close to be a "circle"
numRegions = numel(s);
for k = 1:numRegions
if s(k).Eccentricity < ECCENTRICITY_TH
px = s(k).Centroid(1);
py = s(k).Centroid(2);
ptxt = [' \leftarrow(' num2str(px) ', ' num2str(py) ')'];
text(px, py, ptxt, 'Background', [.5 .5 .5], 'Color', 'blue');
plot(px, py, 'wd', 'MarkerFaceColor', 'blue');
end
end
Of course, you can enhance this piece of code in many ways:
- Make an interactive tool to adjust "gray" threshold (e.g. slider)
- Add an input control to ask for eccentricity threshold
- Allow the user to select the desired "circle" (add 'HitTest', 'on', and 'ButtonDownFcn' callback in the text call. See text and Text Properties)
- Automatically dismiss the perfect circle (the big one -your lenses) that will be the one with the minimum of your eccentricity values (closest to zero)
Hope this help.
