imfindcircles doesn't work

The image looks normal with range from 0 to 4096:

[x,y] = size(AA1); % output x and y both  512

Obviously radius range is [x/4 x/2].

However, both of below:

[centers, radii, metric] = imfindcircles(AA1,[fix(x/4) fix(x/2)],'ObjectPolarity','bright')
[centers, radii, metric] = imfindcircles(AA1,[fix(x/4) fix(x/2)])

output:

centers = []

radii = []

metric = []

changed to [x/6 x/2] doesn't help.

Thank you for finding the issue!

[Image Analyst]

Are you trying to find the mask, which would be the overall circle, or circles inside the circular area? Please indicate what circles you expect to find. Obviously the overall mask could be found by thresholding:

maskBoundary = bwboundaries(AA1 > 0);

If you want circles within that, your data is too blurry and indistinct to find with imfindcircles. You'll have to use a different method to find whatever it is you expect to find in that blurry image.

Also, size() returns [rows, columns] and it's convention to have x be horizontal - the opposite of what you have. You have x as rows (vertical) and y as columns (horizontal dimension) which is the opposite of convention though in this case since they're both 512 it doesn't make any different. Nonetheless, to be less confusing, you should choose descriptive variable names, like

[rows, columns, numberOfColorChannels] = size(grayImage);

[John]

Even made it binary:

AAB = zeros(size(AA1));

AAB(AA1>max(AA1(:))/2) = 1;

figure;imshow(AAB,[])

it showed a great binary circle:

[centers, radii, metric] = imfindcircles(AAB,[for all ranges])

Output still is:

centers = []

radii = []

mtric = []

[Image Analyst]

If you have any more questions, then attach your image and code to read it in with the paperclip icon after you read this:

TUTORIAL: How to ask a question (on Answers) and get a fast answer

Answer 1

[Mathieu NOE]

hello

let's try with your image

AA = imread('circle.png');
AA = rgb2gray(AA);
BW = imbinarize(AA);

[B,L,N] = bwboundaries(BW);
whos

  Name        Size               Bytes  Class      Attributes

  AA        510x505             257550  uint8                
  B           1x1                17768  cell                 
  BW        510x505             257550  logical              
  L         510x505            2060400  double               
  N           1x1                    8  double               
  ans         1x34                  68  char                 

figure;imshow(BW)
hold on 

%Display object boundaries in red and hole boundaries in green.
N

N = 1

for k=1:length(B),
   boundary = B{k};
   if(k > N)
     plot(boundary(:,2), boundary(:,1), 'g','LineWidth',2);
   else
     plot(boundary(:,2), boundary(:,1), 'r','LineWidth',2);     
   end
end

[m,n] =size(BW);
[centers,radii]=imfindcircles(BW,round([n/4,n/2]))

centers = 1×2

252.1513 250.1969

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

radii = 195.0940

[Mathieu NOE]

hello again @John

problem solved ?

[Image Analyst]

I think @John's ignoring us now - he didn't answer my comment up above either.

@Mathieu NOE did you know there is a visboundaries function?

help visboundaries

 visboundaries - Plot region boundaries
    This MATLAB function draws boundaries of regions in the binary image BW
    on the current axes.

    Syntax
      visboundaries(BW)
      visboundaries(B)
      visboundaries(ax,___)
      visboundaries(___,Name,Value)
      h = visboundaries(___)

    Input Arguments
      BW - Binary image
        logical array
      B - Boundary pixel locations
        cell array of Q-by-2 matrices
      ax - Image on which to draw boundaries
        current axes (default) | axes object

    Name-Value Arguments
      Color - Color of boundary
        'red' (default) | RGB triplet | hexadecimal color code |
        color name | short color name
      LineStyle - Style of boundary line
        '-' (default) | '--' | ':' | '-.' | 'none'
      LineWidth - Width of the line used for the boundary
        2 (default) | positive number
      EnhanceVisibility - Augment drawn boundary with contrasting features
        true or 1 (default) | false or 0

    Output Arguments
      h - Boundary lines
        hggroup object

    Examples
      openExample('images/ComputeBoundariesAndPlotOnImageExample')
      openExample('images/VisualizeSegmentationResultExample')

    See also bwboundaries, bwperim, bwtraceboundary, viscircles

    Introduced in Image Processing Toolbox in R2015a
    Documentation for visboundaries
       doc visboundaries

[Mathieu NOE]

hello @Image Analyst

tx for the info, I'll try to remember

as you already know , I am not truly an image expert , that's not my every day job

I'm sometimes trying to improve my basic skills in that field

don't be surprised if my answers are still a bit "basic"

Answer 2

[Matt J]

I can't reproduce the problem you claim to have with a binarized disk. That case is pretty well-handled by imfindcircles:

n=300;
[x,y]=deal(1:n);

mask=hypot(x-mean(x),y'-mean(y))<=n/3;

[centers,radii]=imfindcircles(mask,[n/4,n/2])

centers = 1×2

150.5011 150.5011

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

radii = 99.9962

imshow(mask,[])
viscircles(centers,radii);

●