Hough Transform for circle detectoin - getting information out

[thomas49th]

Hi, I am using the hough transform to perform circle detection. I am able to perform a hough transform and plot the results, but how exactly do I now take my hough transform to actually get meaningful information. I know the size of the my circle (it's radius), but what information am I looking for in the hough transform?

Thanks
Thomas

 

[lanedance]

don't know much about hough transforms, but i assume to detect a circle you'll need to parameterise the likelihood of a circle in terms of a radius and centre point. The highest likelihoods, will be most likely to be a circle (sounds obvious, but hopefully it helps)

can you describe the form of the hough transform?

 

[I like Serena]

Hi, I am using the hough transform to perform circle detection. I am able to perform a hough transform and plot the results, but how exactly do I now take my hough transform to actually get meaningful information. I know the size of the my circle (it's radius), but what information am I looking for in the hough transform?

Thanks
Thomas

You need to make a 2-dimensional histogram of the hough transforms.
The hough transform gives all possible values for the x- and y- values of the center of your circle.
When you put those measurements in a 2-dimensional histogram, somewhere there will be a cell with the highest count. Your circle center will be in this cell.

For a more accurate measurement, you'll need to do more calculations, but the hough transform helps you to find the center of your circle in a picture that contains lots of noise and that may contain only part of your circle.