[Insight-users] membership

[Silvano Agliozzo]

Hi,

I'm trying to implement a clustering algorithm using a KdTree (provided by
itk)  as a data structure. I need to label the 3-d data points that belong
to a certain cluster within a given distance during the running algorithm.
Nevertheless if I use a 4-d measurement vector with this cluster label I
change the distance calculation among vectors as I change their label
values. I wonder if there is a way to organize data point in order to 
carry out the clustering by labeling the point. 

itk::Statistics::MembershipSample provide a similar functionality but in 
order to make it working it requires to know a-priori the number of 
clusters.

thanking you in advance,

Silvano