[Insight-users] Size of kernel in *GaussianImageFilter

[David Doria]

In both SmoothingRecursiveGaussianImageFilter and
DiscreteGaussianImageFilter, there is a single parameter, 'sigma'
(in SmoothingRecursiveGaussianImageFilter) and 'variance'
(in DiscreteGaussianImageFilter). In both cases, this is the variance of
the Gaussian kernel that is to be used. However, I don't see a way to set
the discrete size of the kernel to be used? That is, if you apply a
sigma=10 kernel with a 10x10 kernel, that will produce a very different
result than applying a sigma=10 kernel with a 2x2 kernel. Is this kernel
size automatically set based on sigma/variance?

A related question - with variance=0, the output
of DiscreteGaussianImageFilter is an unmodified version of the input image.
However, with sigma=0, the output of SmoothingRecursiveGaussianImageFilter
is simply an all black image. Is this a bug? Or does it mean that there is
a different interpretation of sigma vs variance in these filters?

Thanks,

David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20120723/1a0da305/attachment.htm>