[Insight-users] Stop flooding pools when "water sheds" would be introduced or use "fill holes" for that

[Luis Ibanez]

Lynx,

Could you post in a public web site a graphical illustration
of the image process that you are looking for ?

Your description of the problem is not yet clear enough for
us to be able to give you a good advice.

It will be useful to see a screenshot of your input image,
and also a diagram of what you would like the output image
to be.


    Thanks


           Luis


-------------------------------------------------------------------
On Wed, Feb 17, 2010 at 3:08 AM,  <lynx.abraxas at freenet.de> wrote:
> On 17/02/10 08:37:54, Richard Beare wrote:
>> I don't think I understand your problem fully, but here are some
>> things you could try.
>
> Thanks  Richard  for  Your  answer.  I  know  about the general way of doing a
> watershed separation. The thing is I have a fully computed image  without  any
> nois.
> And  now I'm looking for a filter result that not only labels but also acts as
> a mask. I have local minima in my image and I need to label  the  reagions  in
> which  each  minimum  is  the  only  and a global one. I want those regions to
> extend as far as the "depth" of the point from which  on  a  pool  of  another
> minimum  would  start flowing into the other. If I'm not mistaken those points
> would be saddle points.
> From my unterstanding of water shed the filter floods each individual pool and
> introduces  water  sheds  as soon as the pools would flood another. I want the
> watershed filter to stop flooding those pools at that point and just keep  the
> so far labeld ragion in which only one definit minimum exists.
>
> Is  it  possible  to  introduce  this flood-stop-condition into the water shed
> filter?
> Or can the same result be achieved with the fill holes filter as  I  discribed
> in my previous mail?
>
> Many thanks for any help or hints about this.
> Lynx
>
>
>
>>
>> It sounds like you are trying to label darker regions. A standard
>> morphological watershed (in review section of ITK, otherwise on IJ)
>> will do this. Watershed labels don't overlap, due to the method of
>> construction. However if you simply apply the watershed to the raw
>> image then you'll get a label for each minima, including all the
>> noise. Usually you'll want to do some preprocessing to reduce the
>> number of minima or find markers somewhere in the basins that you're
>> actually interested in. If you can reliably find such markers then you
>> can use a marker based watershed, which will only produce as many
>> labels as you have markers.
>>
>> On Wed, Feb 17, 2010 at 12:45 AM,  <lynx.abraxas at freenet.de> wrote:
>> > Hello!
>> >
>> >
>> > I'm  looking  for  a  filter  who's result is that of a water shed filter that
>> > would stop filling basins as soon as a water shed  would  be  introduced.  The
>> > result  would  be  a  label  for each marker (eg. local minimum) but the label
>> > would only be as big as the "natural" basin. So the  labels  would  only  just
>> > touch each other.
>> > From  the  description  of the fill-holes-filter I'm not sure if that is doing
>> > what I'm looking for. I tried:
>> > HMinimaImageFilter(float-image)
>> > GrayscaleFillholeImageFilterr(Hminima-result)
>> > SubtractImageFilter(Hminima-result,fillholes-result)
>> > BinaryThresholdImageFilter(Substract-result) #bg==0 fg everything except 0
>> > ConnectedComponentImageFilter(threshold-result)
>> >
>> > But the result is not as expected.
>> > For one I wonder how I can achieve a threshold as fg is everything except 0.
>> > Second the subtraction does not yield 0 any where. Why that? Is this a rouding
>> > error problem?
>> >
>> > Is there a better way to achieve this goal?
>> >
>> > Many thanks for any help or hints
>> > Lynx
>> >
>
>
> _____________________________________
> Powered by 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
>