[Insight-users] canny edge detector behaviour

[Zachary Pincus]

First, I strongly suggest reading up on how precisely the Canny filter 
works. This should give you the tools to diagnose various problems with 
it. See, for example: http://www.cee.hw.ac.uk/hipr/html/canny.html or 
http://www.pages.drexel.edu/~weg22/can_tut.html (These are the first 
two hits for "Canny edge" on Google.)

Now, have you set a threshold for the Canny edge detector? Basically, 
what you're doing with the detector is first smoothing the image, then 
looking for sharp edges. But since you smoothed, there is no longer a 
single sharp edge, but a smooth gradient of edge.

If you don't set a threshold, there will be several concentric edges of 
varying intensities in the smoothed image. So you need either to (a) 
NOT smooth the image first, or smooth it a very small amount, or (b) 
set a reasonable threshold value to prevent the secondary edges from 
being traced out.

Zach Pincus

Department of Biochemistry and Program in Biomedical Informatics
Stanford University School of Medicine


On Jun 29, 2004, at 1:43 PM, David Macias Verde wrote:

> Hi all:
>
> I missed a message with the answer to this question. I didn't look at 
> it
> at the right time but I stored it and thought it was in a save place; 
> it
> wasn't.
>
> So, here it is again. Why is the Canny Edge Detector given more than 
> one
> edge when, for example, I give it a solid ball? The result to this are
> several circles with the same center, depending on the gaussian
> variance. Why is not only 1 even if the gaussian variance is 0?
>
> Thanks,
>
> David
>
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>