| watershed segmentation [message #9217] |
Fri, 06 June 1997 00:00  |
Alrik van den Brom
Messages: 2
Registered: June 1997
|
Junior Member |
|
|
Hi,
does anyone know if there exists an IDL routine for the watershed
segmentation algorithm? It's a algorithm, used in digital image
processing, for separating overlapping features on a binary picture.
Groet,
Alrik.
************************************************************ ******************
Alrik van den Brom
woonachtig op: De Residentie
Hondsrug 184
3524 BV
Utrecht
030-2895597
E-mail: a.j.vandenbrom@fys.ruu.nl or a.j.vandenbrom@chem.ruu.nl
************************************************************ ******************
|
|
|
|
| Re: watershed segmentation [message #9297 is a reply to message #9217] |
Wed, 11 June 1997 00:00  |
John Votaw
Messages: 6
Registered: June 1997
|
Junior Member |
|
|
Karsten Rodenacker wrote:
>
> Alrik van den Brom wrote:
>>
>> Hi,
>>
>> does anyone know if there exists an IDL routine for the watershed
>> segmentation algorithm? It's a algorithm, used in digital image
>> processing, for separating overlapping features on a binary picture.
>>
>> Groet,
>>
>> Alrik.
>>
> Vice versa, I am strongly interested in a 3-D watershed implementation
> as well as a quick 3-D distance transformation!
>
I have an implementation of the 3D watershed operation that I am willing
to share. I should warn you of a few things first. My application is
in PET imaging and I typically have fltarr(128,128,50) image volumns of
rather noisey data. My implementation is NOT a binary implementation.
You give it the image volume and a point of interest. To keep with the
water shed analogy, the water table is initially set to the value of the
point of interest. The water table is then slowly raised (image
threshold reduced) until the basin that contains the original point
merges with an initially distinct basin. Because I have noisey data, I
use a neighborhood operation around the edge of the basin to skip over
basins (small puddles) that are only 1 or a few pixels wide.
Also, I could not figure out how to implement the algorithm without
using nested loops so the code is probably not optimal.
If you would like to try the programs, send me email directly. I would
very much appriciate users that give me comments to improve my programs
and/or programming style.
----
John R. Votaw, Ph.D.
Associate Professor of Radiology
Emory Center for Positron Emission Tomography
votaw@commander.eushc.org
voice: (404)712-7954 FAX: (404)712-7962
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
