[Insight-developers] Problems with Variance in CannyEdgeDetectionImageFilter

[Bradley Lowekamp]

I had a program which used larger values for the variance in the Canny  
edge filter algorithm. As I am updating the program to a newer version  
of the ITK library I had discovered that this is problematic with  
changes made in this algorithm. My observations can be observed with  
the Canny edge example program. With the following variance values I  
get the following:

~1 : good results
~2-4: warning of "itkGaussianOperator.txx", line 54 kernel being  
truncated. Results look plausible
~8+: the out put is an image full of nan (ugly!)

Looking at the CVS, and knowing I was using ITK version 2.8.1 or 3.0  
before. I am really not sure where this problem came from. Perhaps  
there was a change to DiscreteGaussianImageFilter?

Any idea's why this class broke how I was using it?

Also the method CannyEdgeDetectionImageFilter::SetMaximumError  has a  
small mistake in it.

if (m_Variance[i] != v)

should be:

if (m_MaximumError[i] != v)

Thank,
Brad
========================================================
Bradley Lowekamp
Lockheed Martin Contractor for
Office of High Performance Computing and Communications
National Library of Medicine
blowekamp at mail.nih.gov


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/mailman/private/insight-developers/attachments/20080602/0d707b25/attachment.htm>