Improving circle detection with imfindcircle

Hello all, I am trying to improve best-fit circle detection using the function "imfindcircle". Currently a sensitivity of 0.98 searching within a range of [50 175] yield the following:

[centers, radii] = imfindcircles(im{k+1},[50 175],'ObjectPolarity','bright','Sensitivity',0.98)

(This is only a line of the code that is part of a larger loop that performs imfindcircles for multiple images)

I have tried a variety of sensitivities and ranges as well as preprocessing the image by converting it to edges:

imtemp = imread(strcat(X,side,'-',num2str(i1),'.png')); %read
im2{i1+1} = imtemp{i1+1}(:,:,2);
im{i1+1} = edge(im2{i1+1},'canny'); %convert to edges only

However the fits do not seem to improve beyond the original result. What are the best image preprocessing steps to improve detection of the circle?

Thank you!

Answer 1

[Ramnarayan Krishnamurthy]

The following link may have some useful tips on circle detection using imfindcircles and regionprops:

https://www.mathworks.com/matlabcentral/answers/334096-why-does-imfindcircles-not-find-circles-in-my-image?s_tid=answers_rc1-3_p3_MLT

An example is available at:

https://www.mathworks.com/examples/image/mw/images-ex49644526-detect-and-measure-circular-objects-in-an-image?s_tid=answers_rc2-1_p4_MLT