Finding Circles in an Image. Hough Transform, Tao's CircularHough_Grd algorithm, Matlab's infindcircles

Hi

I am trying to find the centre of a ball in an experiment. The image of the ball is well defined, and is easily appreciated by the human eye. I have tried all methods listed above, and even a few more, but the result of finding circles is always nil or some place where the circle isn't present. I am new to Image porcessing, so I don't know what I am doing wrong.

I have tried: 1) edge function 2) Histogram intensity above specific value and many other to process the image before subjecting it to centre finding.

Help!!

The link to the image(after applying adapthisteq to a single color plane): http://postimage.org/image/sahokwd75/

Link to image after applying Tao's algorithm, and plotting the circles http://postimage.org/image/dlktqp3tp/

Answer 1

[Harshit]

Hi Rohan, The intensity of the ball must be near to zero. So Histogram intensity below specific value needed to be taken. Black means zero intensity and white is 255.

Answer 2

[Rohan Budhiraja]

I tried your suggestion. It doesn't help. The algo still picks random circles instead of the desired one

Answer 3

[Image Analyst]

Your image is way too cluttered, especially for a beginner. Try a more uniform background behind the ball.

Answer 4

[Ashish Uthama]

I was able to get the sphere and some false positives with:

 [c, r] = imfindcircles(I, [18 22], 'ObjectPolarity','dark','Sensitivity', .98);
 imshow(I,[]); viscircles(c,r);

See if you can make sense of the parameters from the documentation.