detecting circle using hough tranform

I am trying to detect circle using hough transform using the image above. My code is

RGB = imread('img1.jpg');

imshow(RGB);

Rmin = 60; Rmax = 100;

[center, radius] = imfindcircles(RGB, [Rmin Rmax], 'Sensitivity', 0.9)

viscircles(center, radius);

hold on;

plot(center(:,1), center(:,2), 'yx', 'LineWidth', 2);

hold off;

It's showing error saying

"Index in position 2 exceeds array bounds.

Error in hough (line 7)

plot(center(:,1), center(:,2), 'yx', 'LineWidth', 2);"

Can I get help to fix this? Thanks in advance.

Answer 1

[Simon Chan]

The range of radius is too small and the sensitivity may not be good enough.

I choose radius from 100 to 150 pixels and sensitivity 0.98 as follows:

RGB=imread('https://www.mathworks.com/matlabcentral/answers/uploaded_files/1128355/image.png');
imshow(RGB);
Rmin = 100; Rmax = 150;  
[center, radius] = imfindcircles(RGB, [Rmin Rmax], 'Sensitivity', 0.98) 

center = 2×2

815.0664 742.2900 434.8299 613.4308

radius = 2×1

143.3625 127.1976

viscircles(center, radius);
hold on; 
plot(center(:,1), center(:,2), 'yx', 'LineWidth', 2); 
hold off; 

Answer 2

[Image Analyst]

center is probably empty. To correct, try changimg some of the input parameters. But why use hough/imfindcircles? For the image you showed, you can easily find the circles using the ColorThresholder (which does thresholding) or simply thresholding the green channel.

[Md]

unfortunately I was instructed to use hough/imfindcircles and this is the only option.

[Image Analyst]

Oh, you didn't say it was your homework.

How do I get help on homework questions on MATLAB Answers? - MATLAB Answers - MATLAB Central

I've now tagged it as homework.

So what did you think of @Simon Chan's answer? It seems to work, following along the lines of your code.

Answer 3

[Matt J]

I find that imfindcircles is often hard to parameter-tune. Here's an alternative solution using this circle-fitting tool,

https://www.mathworks.com/matlabcentral/fileexchange/87584-object-oriented-tools-for-fitting-conics-and-quadrics

A=imread('Image.png');

BW=bwareaopen(A(:,:,2)>150,500);
bd=bwboundaries(bwconvhull(BW,'objects'));


clear center radius;

imshow(A);hold on
for i=numel(bd):-1:1

 xy=fliplr(bd{i})';

 fitobj=circularFit(xy);

   center{i}=fitobj.center;
   radius{i}=fitobj.radius;
   
 fitobj.showfit(LineWidth=4);

 end; hold off