[Insight-users] region growing
Luis Ibanez
luis.ibanez at kitware.com
Sat Jun 27 18:12:42 EDT 2009
Hi Arun,
When discussing image segmentation methods, you must include the
description of the problem that you are trying to solve, (images
if possible), the parameters that you have tried, the requirements
of your problem.
Stating that results are "unsatisfactory" doesn't provide any
information, since we lack the context in which such claim could
be evaluated.
Listing the flaws of algorithms will be an exercise in vanity
publishing that belongs better to the many pointless papers that
are published in the irreproducible Journals and Conferences in
our field. The bogus practice of searching for the "best" method,
or even claimiing that one method is "better" than another, is
a common display of the shameless lack of rigor in our field.
A more useful exercise will be for you to describe that problem
that you are trying to solve, and to provide enough information
about your experiments that will allow anybody to replicate them
and therefore, suggest improvements upon them.
Let's start from the beginning:
Please let us know what you are trying to do.
Regards,
Luis
------------------------
Arun Vishwanathan wrote:
> Hi Luis
>
> I have been trying to use connected threshold ,confidence
> connected,isolated connected and neighborhood connected image filter of
> ITK.I dont find the results satisfactory.Could you possibly list what
> could be flaws with the 4 region groiwng algorithms.Are they
> specifically good for some image?or is that the termination criteria or
> the method of deciding the upper and lower threshold is not good?
> regards
> Aaron
>
More information about the Insight-users
mailing list