Cluster analysis code [message #6792] |
Mon, 12 August 1996 00:00  |
Andrea Spinelli
Messages: 4 Registered: August 1996
|
Junior Member |
|
|
Hi everyone,
I am looking for a cluster analysis procedure written, of
course, in IDL. I have browsed several ftp sites, but
I was not able to find anything interesting. Is there
anybody who can help me?
Thanks in advance,
Andrea
--
Andrea Spinelli - Ismes Spa. 9, Via Pastrengo, 24068 Seriate BG, Italy
tel.: +39-35-307777 fax.: +39-35-302999 e-mail: aspinelli@ismes.it
"Truth hurts, but pimples much more"
|
|
|
Re: Cluster analysis [message #11135 is a reply to message #6792] |
Mon, 02 March 1998 00:00  |
Alex Schuster
Messages: 124 Registered: February 1997
|
Senior Member |
|
|
John Dickson wrote:
> I have a 3D image array containing a series of clusters of pixels
> of value 1, in a sea of value zero pixels.
>
> Has anyone an idea on how to produce an algorithm to count the
> number of pixels in each of these clusters?
Are the clusters connected? If not, I would look at the SEARCH3D
routine.
If you need more than that, maybe the routines NN_LEARN and NN_CLUST
could help, this is an implementations of the k-means cluster analysis.
I don't recall where I got them from, they were a bit hard to find, so
if anyone is interested, just ask me and I mail them.
Alex
--
Alex Schuster Wonko@weird.cologne.de PGP Key available
alex@pet.mpin-koeln.mpg.de
|
|
|
Re: Cluster analysis [message #11151 is a reply to message #6792] |
Fri, 27 February 1998 00:00  |
Ewan A. Macpherson
Messages: 14 Registered: December 1996
|
Junior Member |
|
|
John Dickson wrote:
>
> I have a 3D image array containing a series of clusters
> of pixels of value 1, in a sea of value zero pixels.
>
> Has anyone an idea on how to produce an algorithm to count
> the number of pixels in each of these clusters?
Are these contiguous clusters or just regions of higher density? If the
former, you could easily use a minimum spanning tree algorithm to
identify and label the clusters. I have some (inefficient, I fear)
PV-Wave code you'd be welcome to.
--
Ewan Macpherson
|
|
|