quality of the circles found - imfindcircles

Hello,

I am unsing imfindcircles with a greyscale image. Some of the circles identified are good, and some are not so good.

I am now overlapping the circles identified onto the original image and inspect visually.

Does someone know if there is another way of quantifying the quality of the circles found?

Answer 1

[Image Analyst]

The third output of imfindcircles() gives you a metric.

Circle strengths is the relative strengths for the circle centers, returned as a vector. The value at metric(j) corresponds to the circle with radiusradii(j) centered at centers(j,:).

Answer 2

[Marcel Verduyn]

Thanks for this.

Yest, there is some difference in the relative strengths of the droplets identified in the image below. Still, when I continue the iteration which is now set till 60, I find many circles which should not have been identified.

My code is in development and suggestions are very welcome.

close all;
clear all;
clc;
%Input image
img = imread ('Capture.jpg');
%Show input image
figure, imshow(img);

img = rgb2gray(img);
figure, imshow(img);

[BW,maskedImage,metric] = segmentImage(img);
figure, imshow(maskedImage);

metric(1:10)
figure, histogram(metric)

stats = regionprops('table',BW, 'Centroid',...
      'MajorAxisLength','MinorAxisLength');
  centers = stats.Centroid;
  diameters =mean([stats.MajorAxisLength stats.MinorAxisLength],2);
  radii = diameters/2;
  
  figure, imshow(img)
  hold on
  viscircles(centers,radii)
  hold off

function [BW,maskedImage,metric] = segmentImage(X)
%segmentImage Segment image using auto-generated code from imageSegmenter App
%  [BW,MASKEDIMAGE] = segmentImage(X) segments image X using auto-generated
%  code from the imageSegmenter App. The final segmentation is returned in
%  BW, and a masked image is returned in MASKEDIMAGE.

% Auto-generated by imageSegmenter app on 14-Nov-2018
%----------------------------------------------------


% Find circles
[centers,radii,metric] = imfindcircles(X,[63 88],'ObjectPolarity','dark','Sensitivity',1);
BW = false(size(X,1),size(X,2));
[Xgrid,Ygrid] = meshgrid(1:size(BW,2),1:size(BW,1));
for n = 1:60
    BW = BW | (hypot(Xgrid-centers(n,1),Ygrid-centers(n,2)) <= radii(n));
end;

% Create masked image.
maskedImage = X;
maskedImage(~BW) = 0;
end

[Image Analyst]

You can try the method in the attached paper.

Answer 3

[Marcel Verduyn]

Is this method implemented in Matlab?

I'd first like to try to improve the script I am using now because imfindcircles finds 8 large circles, of which only 1 does not match. Only when I increase the number of iteration steps (higher than 60) more unmatching circles are found.