[Insight-users] muliple pahse Chan Vese, one partition totally shrink

[Arnaud GELAS]

Just a comment regarding the kd-tree optimization, this will work only 
if considered objects are really distinct. (One object must not contain 
another one).
But if I correctly understood what you want to do, this should not apply 
to your case...

On 03/26/2010 01:25 PM, Kishore Mosaliganti wrote:
> Hi Baoyun,
>
> You are right that if you set 4 phases in the levelset, then no phase 
> will shrink to empty.
>
> In ITK, if you use the classes on 
> ScalarChanAndVeseLevelSetImageFilter, you can separately encode the 3 
> regions in 3 separate level-set functions and these functions compete 
> with each other for the segmentations at the boundaries. You can also 
> control each level-set parameters separately.
>
> If you have N objects, then you can make use of a built-in kd-tree 
> optimization that speeds up the approach. Each levelset function is 
> only defined in a small subdomain containing your object instead of 
> the whole image thereby saving space.
>
>
> Kishore
>
> On Fri, Mar 26, 2010 at 10:48 AM, Baoyun Li <baoyun_li123 at yahoo.com 
> <mailto:baoyun_li123 at yahoo.com>> wrote:
>
>     Dear All:
>     Thanks for your reply, now seems I undertand the question more.
>     According to Wanlin's paper, I need to encoding the 2 level set
>     to  3 phase if I want to 3 segmenations. If I need 4 
>     segmentation, I need to encode the levelset to 4 phases.
>     So my origianl question is not correct, once I set 4 phases, no
>     phase will shrink to empty, and I will get 4 segmentation anyway.
>     (Please correct me if I am wrong).
>     Thanks and my best regards
>     Baoyun
>     Please correct me if I am wrong
>
>     --- On *Thu, 3/25/10, Luis Ibanez /<luis.ibanez at kitware.com
>     <mailto:luis.ibanez at kitware.com>>/* wrote:
>
>
>         From: Luis Ibanez <luis.ibanez at kitware.com
>         <mailto:luis.ibanez at kitware.com>>
>
>         Subject: Re: [Insight-users] muliple pahse Chan Vese, one
>         partition totally shrink
>         To: "Baoyun Li" <baoyun_li123 at yahoo.com
>         <mailto:baoyun_li123 at yahoo.com>>
>         Cc: insight-users at itk.org <mailto:insight-users at itk.org>,
>         "Mosaliganti, Kishore Rao"
>         <Kishore_Mosaliganti at hms.harvard.edu
>         <mailto:Kishore_Mosaliganti at hms.harvard.edu>>
>         Date: Thursday, March 25, 2010, 10:43 PM
>
>
>
>         Hi Baoyun,
>
>
>         You seem to be mixing two different issues here.
>
>         One thing is to segment the image in different regions,
>         and a separate issue is to find an efficient encoding
>         for the resulting segmentations.
>
>
>         If you want to segment three different anatomical structures
>         from your image, then you want to use a level set with
>         three phases.
>
>         That is, one phase per object to be segmented.
>
>
>               Regards,
>
>
>                     Luis
>
>
>         -----------------------------------------------------------------------------
>         On Wed, Mar 24, 2010 at 5:52 PM, Baoyun Li
>         <baoyun_li123 at yahoo.com
>         <http://us.mc1110.mail.yahoo.com/mc/compose?to=baoyun_li123@yahoo.com>>
>         wrote:
>
>             Dear All:
>             I have a theoritical question about Chen Vese multiplse
>             pahse leve set.
>             I have two level set and the two level set can segment the
>             image to 4 segments, then I can calculate the mean for
>             each segments as following
>             %c11 = mean (phi1>0 & phi2>0)
>             %c12 = mean (phi1>0 & phi2<0)
>             %c21 = mean (phi1<0 & phi2>0)
>             %c22 = mean (phi1<0 & phi2<0)
>             How about if I only have three segments, let say segment
>             (phi1<0 & phi2<0) will finally shrink to zeros.
>             Then how to I calcuate c22, and also what value of
>             c22 I should use to update leve set equation as shown below:
>             ((u-c22)^2-(u-c12)^2)H(phi2).
>             I saw somebody use the mininum number of the machine to
>             replace c22, but I could not imagine what will happen and
>             why to do that.
>             Thanks
>             Baoyun
>
>
>
>             _____________________________________
>             Powered by www.kitware.com <http://www.kitware.com/>
>
>             Visit other Kitware open-source projects at
>             http://www.kitware.com/opensource/opensource.html
>
>             Kitware offers ITK Training Courses, for more information
>             visit:
>             http://www.kitware.com/products/protraining.html
>
>             Please keep messages on-topic and check the ITK FAQ at:
>             http://www.itk.org/Wiki/ITK_FAQ
>
>             Follow this link to subscribe/unsubscribe:
>             http://www.itk.org/mailman/listinfo/insight-users
>
>
>
>
>     _____________________________________
>     Powered by www.kitware.com <http://www.kitware.com>
>
>     Visit other Kitware open-source projects at
>     http://www.kitware.com/opensource/opensource.html
>
>     Kitware offers ITK Training Courses, for more information visit:
>     http://www.kitware.com/products/protraining.html
>
>     Please keep messages on-topic and check the ITK FAQ at:
>     http://www.itk.org/Wiki/ITK_FAQ
>
>     Follow this link to subscribe/unsubscribe:
>     http://www.itk.org/mailman/listinfo/insight-users
>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20100326/b6a0787f/attachment-0001.htm>