| Re: Thinning image morphological operator [message #49233 is a reply to message #49160] |
Wed, 05 July 2006 07:21  |
Ulan
Messages: 8
Registered: April 2005
|
Junior Member |
|
|
Yeah, I 've also noticed connectivity loss with THIN. I used Laplacian
filter prior to thinning to insure the THIN doesn't remove thinner
parts ...
Regards,
Ulan
Karsten Rodenacker wrote:
> Ja, you are right. The idl implementation THIN seems to result from times
> where connectivity was an unfamiliar term. The routine is comparingly
> quick but not connectivity preserving. Maybe a closing (MORPH_CLOSE) might
> help in some cases.
>
> Unluckily the implementation of EROSION and DILATION is likewise not just
> satisfying. I had a long fruitless discussion about that with RSI. At
> least they add some more documentation concerning the unusual border
> behaviour. Possibly critical remarks might help to convince VTT that there
> is some necessity to train somebody in mathematical morphology. In terms
> of math. morph and connectivity my reference is a very old program from
> Ecole de Mines, Fontainebleau, France, microMORPH, unluckily not freeware,
> where connectivity and border behaviour are implemented consistently.
>
> Regards
> Karsten
>
> Am Fri, 30 Jun 2006 00:15:04 +0200 schrieb Tom S. <twslankard@gmail.com>:
>
>> I had a question about the THIN function in IDL. I was under the
>> impression that this operation was supposed to preserve connectivity.
>> Am I mistaken? The IDL implementation does not seem to do so.
>>
>> The image linked below is output from a program I wrote. It simply
>> takes the image on the left and applies the THIN function, yielding the
>> image on the right. (Note the gap that forms toward the bottom.)
>>
>> http://twslankard.googlepages.com/fluke.jpg
>>
>> Is this a problem with my understanding of the thinning operation or is
>> it a problem with the THIN function? Any assistance is greatly
>> appreciated!
>>
>> Regards,
>> Tom S.
>>
>
>
>
> --
> Erstellt mit Operas revolutionärem E-Mail-Modul: http://www.opera.com/m2/
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
